Module 1: The 10x Intern Mindset
What an internship is really for, and where AI fits
Why an internship is a short window to collect skills, trust, a reference, a portfolio and direction, how LI works in Malaysia, what supervisors value most, and a free starter AI toolkit you can use with care.
Learning Objectives - Explain what an internship is for beyond completing tasks
- Describe how LI works in Malaysia, including the two supervisors, the logbook and assessment
- Apply the habits supervisors value most: reliability, good questions and honesty
- Distinguish what AI can speed up from what remains your responsibility
- Set up a free starter AI toolkit and use it with privacy in mind
What You'll Learn - Skills, trust, a reference, a portfolio and direction
- What 10x really means
- LI, faculty and industry supervisors, and the logbook
- TalentCorp and MySIP
- Reliability, good questions and confirming the brief
- What AI changes and what it does not
- Free AI tools and their limits
- Free does not mean private
What an Internship Is Actually For
An internship looks like a short job. You get a desk, or a corner of one, a lanyard, some tasks and a supervisor. But if you treat it only as a job, you will measure yourself by output: how many tasks you finished and how busy you looked. That is the wrong scoreboard. A company does not take on an undergraduate for ten or twenty weeks because it needs cheap output. It takes you on to see what you might become, and you go in to find out what the work is really like.
Five things worth leaving with
Think of the internship as a short window to collect five things. The first is
skills: the practical, unglamorous ones that lectures rarely teach, such as writing an email a busy manager will actually read, formatting a report to a house style, or finding a file on a shared drive nobody labelled properly. The second is
trust: the moment your supervisor stops checking every line you send. The third is a
reference, a named person who will vouch for you when a future employer calls. The fourth is a
portfolio of real work you are allowed to show. The fifth is
direction: a clearer sense of whether this field, this kind of company and this kind of working day suit you. Finding out that you do not want to work in an agency is a useful result, not a failure.
Five things to leave with - skills, trust, a reference, a portfolio and direction
What 10x really means
This course is called 10x Internship, but the 10x is not about producing ten times as much as other interns. Nobody wants an intern who floods the team with half-checked drafts. It means a
steeper learning curve: reaching in week three the understanding most interns reach in week ten, because you prepared, asked better questions and used AI to learn faster. Output follows understanding, not the other way round.
AI tools fit into this picture as accelerators. They can explain an unfamiliar term in seconds, help you practise a difficult conversation, or turn your rough notes into a first draft. They cannot earn trust for you, and they cannot give you a reference. Those come from how you behave: whether you are reliable, honest about what you do not know, and careful with what the company shares with you. The rest of this course is about combining both halves.
Watch video: What an Internship Is Actually For
Key Insight: Measure your internship by what you leave with - skills, trust, a reference, a portfolio and direction - not by how busy you looked. In this course, 10x means learning faster, not producing more.
Real-World Example: Before her first day at Teras Komunikasi, a PR agency in Petaling Jaya, Aina writes one line for each of the five things. Skills: learn to write a press release the agency would actually send. Trust: be given a client email to draft without a rewrite. Reference: earn one from her supervisor. Portfolio: three pieces she may show, with permission. Direction: decide whether agency life suits her. She rereads the list every Friday.
Q: In this course, what does the "10x" in 10x Internship mean?
The 10x is about a steeper learning curve: understanding in week three what most interns grasp in week ten. Flooding a team with half-checked drafts does the opposite, because output should follow understanding.
Write your own five lines, one each for skills, trust, a reference, a portfolio and direction, for the internship you are hoping for. Which of the five feels hardest to get, and why?
How Malaysian Internships Work
In most Malaysian universities an internship is not an optional extra. It is
latihan industri, usually shortened to
LI (industrial training): a supervised placement built into your degree, often a credit-bearing course you must pass to graduate. Programmes are accredited by the
Malaysian Qualifications Agency (MQA), and each university and faculty sets its own LI rules. That is why your friends in engineering and in mass communication may have quite different requirements.
How long, and when
There is no single national length. University handbooks show placements from about 9 or 10 weeks up to 24 weeks, and some programmes ask for four to six months. Some degrees go further: under the Ministry of Higher Education's
2u2i work-based learning mode, students spend about two years at university and two in industry, with variants such as 3u1i. Your own programme structure and your faculty's LI handbook are the sources that count for you.
Two supervisors, one logbook
You will normally have two supervisors. Your
industry supervisor, sometimes called the organisation or company supervisor, gives you work at the company and watches how you do it. Your
faculty supervisor is a lecturer who checks on your progress, may visit or call your workplace, and marks part of your assessment. The two may rarely speak, so you are the link between them.
Most programmes also ask for a
logbook: a daily or weekly record of what you did, often signed and stamped by your industry supervisor. Faculties take it seriously. Some give it a sizeable share of the marks, and some penalise a logbook that is late or unsigned. Assessment usually combines the two supervisors' evaluations, the logbook and a final report or presentation, with weightings that differ by faculty.
Where TalentCorp fits
Talent Corporation Malaysia (TalentCorp), working under the Ministry of Human Resources, runs schemes that encourage companies to take interns. Under the
National Structured Internship Programme (MySIP), an employer whose internship is endorsed by TalentCorp can claim a double tax deduction on intern expenses, provided the placement lasts at least 10 weeks and pays at least RM600 a month for degree students, or RM500 at diploma level. No general law sets a minimum allowance for every intern, so allowances vary widely. Ask about the allowance openly before you accept a place.
Key Insight: Your faculty's LI handbook is your rulebook. Length, logbook format, forms and assessment weightings differ between universities and even between faculties, so read yours before you start applying.
Real-World Example: Hafiz, an accounting student, reads his faculty's LI handbook a semester early. He learns that his placement must last 24 weeks, that his logbook needs his industry supervisor's signature every week rather than once at the end, and that the company's evaluation form goes straight to his faculty. On his first day at the audit firm in Johor Bahru he asks his supervisor which afternoon each week suits her for signing.
Q: According to this section, which source decides how long your LI must last and how it is assessed?
MQA accredits programmes, but each university and faculty sets its own LI rules, so lengths range from about 9 weeks to 24 weeks or more. MySIP sets conditions for employers claiming a tax deduction, not for your degree.
Action step: find your own faculty's LI handbook or briefing slides this week. Note three things: how long your placement must be, who signs your logbook and how often, and how the marks are split. Share what you found and I will help you turn it into a simple checklist.
What Supervisors Wish Interns Knew
Your industry supervisor probably did not ask for an intern. In many companies the role is added on top of a full workload, with no reduction in deadlines. That is not a reason to feel like a burden. It is a reason to make supervising you easy. Ask supervisors what they wish interns knew and the same few points come up again and again, and almost none of them are about talent.
Reliable beats brilliant
A supervisor would rather have an intern who does a small task completely and on time than one who does a dazzling job on half of it. Completely means checked, named properly, saved where the team can find it and followed up. When you finish, say so. A one-line message such as "The media list is done and saved in the shared folder" closes the loop, so nobody has to chase you. Reliability is how trust starts.
Ask well, and early
Supervisors want questions. What frustrates them is the question that comes too late, on the day of the deadline, or comes empty, such as "I don't understand". A good question shows what you tried, says exactly what is blocking you and says when you need an answer. If you have several small questions, collect them and ask in one go rather than interrupting five times. If a task will be late, say so as soon as you know, not when it is due. Bad news early is a favour; bad news late is a problem.
Confirm the brief
Many intern mistakes start in the first two minutes, when a task is explained quickly and the intern simply nods. Repeat the task back in your own words: what it is for, what the finished piece looks like and when it is due. It feels awkward the first time, and it saves hours.
Do not bluff
Saying "I'm not sure, let me find out" is always safer than guessing. This matters even more now that AI can produce a confident answer to anything. A supervisor who discovers that you passed off a guess, or an unchecked AI answer, as fact will check everything you do from then on. A supervisor who hears an honest "I don't know yet" learns that your yes means yes.
Finally, take notes. Being told something twice is normal. Being told three times suggests you were not listening.
Watch video: What Supervisors Wish Interns Knew
Key Insight: Supervisors rarely ask for brilliance. They ask for small tasks done completely, questions asked early and well, briefs confirmed before you start, and honesty when you do not know.
Real-World Example: Mei Ling, at a fintech start-up in Cyberjaya, has spent an hour stuck on a data clean-up script. Instead of "It's not working", she messages her supervisor: "I'm cleaning the March transactions file. Dates in two formats are breaking the import. I tried converting them in the script, but about 40 rows still fail. Could you spare ten minutes before 3pm? The report is due tomorrow."
Q: You have been stuck on a task for an hour. Which message to your supervisor best follows this section's advice?
A good question shows what you tried, names the exact blocker and gives a time. An empty question pushes all the thinking back to your supervisor, and waiting until the deadline turns a small problem into a late one.
Role-play: I will be your industry supervisor on your first morning. I am busy, and I will give you a task quickly, in one breath, and then turn back to my screen. Your job is to confirm the brief: repeat back what it is for, what the finished piece looks like and when it is due. Tell me what field you are interning in, and I will give you the task.
What AI Changes, and What It Doesn't
A few years ago, an intern's first weeks were mostly spent being slow: slow to understand the jargon, slow to write a first draft, slow to find out how things are done. Generative AI tools change the speed of much of that. They do not change what the internship is for.
What AI changes
AI is very good at the parts of learning that used to need a patient colleague. You can ask it to explain an unfamiliar term, such as "What is a media advisory, and how is it different from a press release?", as many times as you need without feeling foolish. You can paste in your own rough notes and get a tidier draft to improve. You can ask it to quiz you, play a difficult client so you can rehearse, or suggest several ways to structure a report. For a language or communication student, it can offer alternative phrasings and explain why one sounds more formal than another. Used like this, AI shortens the distance between being confused and being useful.
What AI changes and what it does not - speed on one side, responsibility on the other
What it does not change
AI does not know your company. It has not read the client's brief, the house style guide or last week's email thread, and it does not know which facts are confidential. It can be fluently, confidently wrong, inventing a statistic, a quote or a source that looks perfectly real. And it cannot take responsibility. When a press release goes out with the wrong figure, nobody blames the tool; they ask who sent it.
So the things that earn trust stay with you:
judgement about what suits this audience,
accuracy you have checked yourself,
discretion about what can be shared, and
relationships with the people you work with. These are exactly what a supervisor is assessing.
Accelerator, not substitute
A useful test is whether you could explain and defend every line you hand in without the tool open. If yes, AI helped you learn and work faster. If no, it did the thinking for you, and the gap will show the first time someone asks a follow-up question. Your company may also have rules about which tools you may use and what you may put into them. Find out before you use AI on real work, and follow those rules even when other interns do not.
Key Insight: AI speeds up explaining, drafting and practising. Judgement, accuracy, discretion and relationships stay with you, and so does responsibility for anything you send.
Real-World Example: Darren, a translation student at Lingua Rakan in Kuala Lumpur, asks an AI tool to explain the difference between localisation and transcreation, then asks it to quiz him until he can explain both without notes. When a client job comes in, he does not ask the tool which terms the client prefers. That answer lives in the firm's own glossary, so he asks his supervisor where to find it.
Q: Which test does this section suggest for telling whether AI helped you or did the thinking for you?
If you can explain and defend every line yourself, AI accelerated your learning. If you cannot, the gap shows as soon as someone asks a follow-up question, and that is when trust is lost.
Do you agree that AI changes the speed of an internship but not its purpose? Think of one task in your field where AI could save you real time, and one where relying on it would be risky. Tell me both and we can test where the line sits.
Your Starter AI Toolkit
You do not need to pay for anything to start. Every tool below has a free version that is enough for an intern. Free limits change so often that the safest rule is to check each tool's own plan page rather than trust any list, including this one.
General assistants
Four chat assistants cover most needs:
ChatGPT from OpenAI,
Gemini from Google,
Claude from Anthropic and
Microsoft Copilot. All of them let you ask questions, draft and rewrite text, and summarise documents you upload. Their free tiers usually cap how much you can do in a given period, or switch you to a smaller model when you reach the cap, and extras such as file uploads and image generation often have their own limits. Each has a different feel, so try the same task in two of them and compare. Check the age rules too: Claude, for example, requires users to be at least 18.
A research notebook
Google's
Gemini Notebook, called NotebookLM until July 2026, works differently. You upload your own sources, such as a company's annual report, a few of its press releases and an industry article, and it bases its answers on those sources, with citations pointing back to the passages it used. That makes it a strong tool for learning a company or an industry quickly, because you can check every answer against the original.
Language tools
Grammarly checks English spelling, grammar and punctuation for free, with a small monthly allowance of AI rewriting on the free plan. For translation,
Google Translate covers Malay, Chinese, Tamil and many other languages, and
DeepL is often preferred for natural phrasing; it handles Chinese and now lists Malay among its languages. For a translation student, these are tools whose output you post-edit, never final answers.
Free does not mean private
On personal free accounts, your conversations may be used to improve the provider's models unless you switch this off. ChatGPT, for instance, has a setting under Data Controls called "Improve the model for everyone". Turning it off is sensible, but it does not make a public tool a safe place for client names, unpublished figures or anything marked confidential. Treat every free tool as public, and check your company's AI policy before you paste in anything from work.
Watch video: Your Starter AI Toolkit
Key Insight: One general assistant, one research notebook and one or two language tools are enough to start. Check each tool's current free limits on its own site, and treat every free tool as public.
Real-World Example: Aina sets up her toolkit the weekend before her internship. She picks one assistant and switches off the setting that lets her chats train its models. She creates a notebook with five of Teras Komunikasi's published press releases and asks it to describe the agency's writing style, checking each answer against the citations. She installs Grammarly for her English drafts. Client files stay out of all three.
Q: Aina has switched off the setting that lets her assistant train on her chats. What does this section say about pasting a client's unpublished figures into it?
Switching off model training is sensible, but it does not turn a public tool into a safe place for confidential material. Treat every free tool as public and check the company's AI policy before using anything from work.
Action step: before your next class, set up one general assistant and one research notebook. In the notebook, add three public documents about a company you would like to intern with, and ask it what the company does and who its customers are. Come back and tell me one thing it got right and one thing you had to check.
Module 2: Landing the Right Placement
Choosing well and applying with AI, without sounding like AI
Choose a placement that will actually teach you, research a company in one evening, write a resume and cover letter that sound like you, build a small portfolio, rehearse interviews with AI, and spot the scams that target students.
Learning Objectives - Judge an internship offer on learning, real work, supervision and practicality
- Research a company in an evening and turn it into a one-page brief
- Use AI to tighten a resume and cover letter without losing your own voice
- Prepare STAR stories and rehearse interviews with an AI interviewer
- Recognise fake job offers and verify an employer before sharing personal data
What You'll Learn - Four questions for choosing a placement
- Where Malaysian students find internships
- A one-evening company research routine
- The You-AI-You drafting method
- Writing samples and a simple portfolio
- The STAR method and AI mock interviews
- Job scam warning signs and how to check an employer
Choosing Where to Intern
Where you intern shapes what you learn, who will vouch for you afterwards and what you can show your next employer. A famous name helps, but a placement where someone teaches you and gives you real work is usually worth more than a well-known logo on your lanyard.
Four questions for any placement
Before you apply, ask four things. Will I
learn skills that match the career I want? Will I
do real work I can show later, not just filing and photocopying? Will someone
supervise me and give feedback? And does it
work in practice: do the dates fit my faculty's LI window, is the commute or move realistic, and will any allowance cover my costs? LI, short for
latihan industri, is the industrial training most Malaysian degree programmes require, so your faculty's rules on duration and approved organisations always come first.
Four Questions for Any Placement - Learn, Do, Supervision, Practical
Sectors for every field
Languages and communication students have more options than they often think. Newsrooms and digital media teach speed and accuracy. PR and corporate communications agencies teach press releases, media relations and client service. Publishers teach editing and production, and broadcasters teach scripting and writing for the ear. Translation and localisation firms teach terminology, subtitling and post-editing, which means correcting machine translation. In-house communications teams in banks, government-linked companies and NGOs teach how a large organisation speaks with one voice. Students in other fields can put the same four questions to audit firms, engineering consultancies, laboratories, start-ups and the public sector.
Where to find placements
Start with your university's LI or career unit, which keeps lists of approved employers and past placements. Then search
LinkedIn,
Jobstreet and
Hiredly, and register on
MYNext, the TalentCorp platform that matches students with internships, including those under the National Structured Internship Programme (MySIP). Seniors who interned last year are an underrated source: ask them what they actually did all day.
AI can help you compare options. Paste three internship adverts into a chatbot and ask it to set out the likely tasks, the skills you would gain and the questions you should ask about each. Treat its answer as a list to check, not a verdict.
Key Insight: Judge a placement on four questions: will you learn, will you do real work, will someone supervise you, and does it work in practice? Start with your university's LI unit, then LinkedIn, Jobstreet, Hiredly and MYNext.
Real-World Example: Aina had two offers. A large multinational's marketing team wanted an intern mainly to enter data into its customer system. Teras Komunikasi, a smaller PR agency in Petaling Jaya, promised press release drafts and media lists under a senior account executive. She chose Teras because it scored on learn, do and supervision, and its office was a short train ride from home.
Q: Aina is choosing between a famous company offering mostly data entry and a smaller agency offering supervised drafting work. What does the section suggest?
The four questions are learn, do, supervision and practicality. A smaller firm that scores well on all four often teaches more than a big name where the work is mostly routine.
Action step: list two or three placements you are considering and score each from 1 to 5 on learn, do, supervision and practicality. Share your scores here and I will help you spot what you still need to find out.
Researching a Company in an Evening
Interviewers often open with "So, what do you know about us?" Most interns answer with the first line of the company website. You can do much better in one focused evening of about two hours.
The sources, in order
Start with the company's own website: what it does, who its clients or customers are, its services and its news page. Next, search for recent news coverage from the past year, which tells you what the company is proud of and what problems it faces. Then read its LinkedIn page and the profiles of a few staff, especially the team you would join, to see their backgrounds and the work they share. Look at its social media accounts to hear how it speaks to the public. If it is a listed company, its latest annual report explains its strategy in its own words.
For language and communication roles, pay special attention to the
voice. Is the tone formal or friendly? Does it publish in English, Bahasa Melayu, Chinese or all three? A PR agency's own press releases show you the style it will expect from you.
Where AI helps, and where it fails
AI is useful for organising what you find. You can paste the text of the About page and three news articles into a chatbot and ask it to summarise the company's main services, recent projects and possible challenges. A tool such as Gemini Notebook (formerly NotebookLM), which answers only from the sources you upload, makes this easier to check.
But be careful asking a chatbot about a company from memory. Many Malaysian firms, especially small agencies and start-ups, have little online coverage, and a chatbot may confidently invent clients, awards or founders that do not exist. Repeating one of those in an interview is worse than saying nothing. Use AI on sources you have read, and check every fact it gives you against the original.
The one-page brief
Finish the evening with a one-page brief in your own words: what the company does, who it serves, two recent pieces of news, how it sounds, three thoughtful questions to ask, and one small idea you could contribute. That page is your preparation for the cover letter and the interview.
Key Insight: Research in this order: company website, recent news, LinkedIn page and staff profiles, social media, and the annual report if listed. Use AI to summarise sources you have read, never to recall facts about a small company from memory.
Real-World Example: Darren researched Lingua Rakan before applying. He read its service pages, found a news story about a subtitling project for a streaming platform, and noticed on LinkedIn that its project managers posted about Malay and Chinese terminology. When a chatbot claimed the firm had won a national translation award, he searched and found no record of it, so he left it out.
Q: Why should you avoid asking a chatbot to describe a small Malaysian company from memory?
Small firms often have little online coverage, so a chatbot may fill the gaps with confident fabrications. Use AI to summarise sources you have actually read, and check every fact against the original.
Role-play: I will be the hiring manager at the company you most want to join. I open the interview with, "Thanks for coming in. Before we start, what do you know about us?" Answer as you would in the room, and I will respond in character before giving you feedback.
A Resume and Cover Letter That Sound Like You
Employers read many applications that all sound the same: "a passionate, dynamic self-starter eager to thrive in a fast-paced environment." Much of that sameness now comes from AI. Your goal is to use AI to sharpen your application while keeping it unmistakably yours.
What an internship resume needs
For an undergraduate, one page, or two at most, is enough. Put your contact details and education first, then relevant coursework and projects, then experience, which can include part-time jobs, campus clubs, volunteering and freelance work. Add skills and languages, and describe your language levels honestly. Write each experience line as an
action, what you did, and a result: "Edited a 24-page faculty newsletter, cutting turnaround from three weeks to ten days" says far more than "Responsible for newsletter."
Tailor the resume to each advert. Many employers use an
applicant tracking system, software that sorts applications, sometimes by matching the words in the advert. Use the advert's own terms where they truthfully describe what you have done.
The cover letter
Keep it to three or four short paragraphs: why this company, using a specific detail from your research; what you offer, with one or two pieces of evidence; and a polite close saying when you are available. Address it to a named person if the advert or the company's LinkedIn page gives one, and mention your LI dates so the employer knows at once whether you fit. Read it aloud before sending: if a sentence sounds like something you would never say, rewrite it until it does.
The You-AI-You Method - Your Facts, AI Polish, Your Final Voice
Using AI without sounding like AI
Use the
You-AI-You method. First, write the raw facts yourself. Then ask AI to suggest a structure, stronger verbs and tighter wording. Finally, rewrite the result in your own voice, removing stock phrases and checking that every claim is true. Never let AI add a skill, a result or a responsibility you did not have: anything on your resume can be the subject of an interview question.
Watch video: A Resume and Cover Letter That Sound Like You
Key Insight: Write experience lines as action, task and result, and tailor each application to the advert. Use the You-AI-You method: your raw facts first, AI for structure and wording, then your own final rewrite with every claim checked.
Real-World Example: Mei Ling's first draft said "Helped with app project." She told a chatbot what she had actually done, and it suggested "Spearheaded a cutting-edge mobile solution." She rejected that as inflated and wrote her own version: "Built the login and payment screens for a four-person campus food-ordering app used by about 300 students."
Q: In the You-AI-You method, what is your job in the final step?
The last step belongs to you: remove stock phrases, restore your own voice and confirm every line is accurate. AI may polish your facts, but it must never invent them.
Paste one line from your resume or cover letter here. Tell me what you actually did and any number you can attach to it, and we will rework it together using action, task and result, keeping it in your voice.
Writing Samples and a Portfolio
A resume tells an employer what you claim you can do. A portfolio shows them. For language and communication roles, a newsroom, agency or translation firm may ask for writing samples or set you a short writing test, and a ready portfolio puts you ahead of applicants who scramble to find something.
Choosing what goes in
Pick three to five of your best pieces, chosen to match the roles you want. A PR applicant might include a press release, a set of social media posts and a feature article. A journalism student might include a news story and an interview piece. A translation student might show a short source text beside their translation, with a note on the difficult terms. Class assignments count, as long as you say what the brief was.
Include only work that is yours to share. Do not include anything confidential from a part-time job, and do not present group work as solely your own: name your role and credit your teammates. If a piece was published, link to it or note where it appeared.
If you have nothing yet
Create samples. Write a mock press release for a real public event, a short explainer about a topic in your field, or a translation of a public notice. Mark these clearly as practice pieces. Employers would rather see an honest, well-made practice piece than nothing at all.
Presenting it
A tidy PDF or a simple free website is enough. Give each sample a two or three line note: the brief or audience, your role, and the tools you used. Be honest about AI here. If AI helped you brainstorm or check grammar, say so briefly. A portfolio is meant to show your own writing, so the words should be yours, and you should be able to talk through every choice you made.
Students in other fields can build the same habit. A computer science student can share a small project with clear notes on what they built. An accounting student can show a spreadsheet model built on made-up data. An engineering student can share a project report summary, with anything sensitive removed.
Key Insight: Choose three to five pieces that match the roles you want, include only work that is yours to share, credit teammates, and add a short note on the brief, your role and the tools you used, including any AI help.
Real-World Example: Darren built a four-piece portfolio: a Malay-to-English translation of a public council notice set side by side, a short English-Chinese glossary he made for a class project, subtitles for a two-minute public information video, and a note on where he had post-edited machine output. Each piece had a two-line note on the brief and his method.
Q: You wrote a feature article as part of a three-person class project. How should it appear in your portfolio?
Group work can go in, as long as you are honest about it. State your role and credit your teammates, so an employer knows exactly what you can do on your own.
Do you agree that an honest practice piece is better than no sample at all? Tell me which roles you are applying for, and I will help you plan three portfolio pieces you could finish in the next two weeks.
Mock Interviews with AI
Interviews reward practice, and AI gives you a patient interviewer at any hour. Used well, it helps you go into the real interview with your stories ready and your nerves under control.
The questions you will meet
Most internship interviews mix four kinds of question: about you ("Tell me about yourself"), about them ("Why do you want to intern here?"),
behavioural questions that ask for a past example ("Tell me about a time you worked to a tight deadline"), and sometimes a practical task, such as editing a paragraph or translating a short text on the spot.
The STAR method
For behavioural questions, use
STAR:
Situation (the context, briefly),
Task (what you had to achieve),
Action (what you did) and
Result (what happened, with a number if you have one, and what you learned). Spend most of your time on the Action, and say "I" rather than "we", because the interviewer wants to know what you did. Prepare five or six stories from your studies, part-time work, clubs or volunteering. One good story can often answer several different questions.
Turning AI into an interviewer
Give a chatbot the job advert and your resume, and ask it to interview you for that role, asking one question at a time and waiting for your answer. After each answer, ask for feedback on structure, length and whether you used STAR. Ask it to add follow-up questions, because real interviewers dig deeper. If your tool has a voice mode, practise out loud, since an answer that reads well can sound very different when spoken. If the interview may be in Bahasa Melayu, practise in Malay too.
Know the limits
AI feedback can be generic, and a chatbot cannot judge your eye contact or how you come across in a room. Do at least one practice with a real person, such as a lecturer, a career officer or a friend. And never use AI to feed you answers during a real interview, including an online one. It is dishonest, it is usually noticeable, and the point of the interview is to meet you.
Finally, prepare two or three questions of your own, drawn from your one-page brief.
Watch video: Mock Interviews with AI
Key Insight: Answer behavioural questions with STAR: Situation, Task, Action, Result, spending most time on your own actions. Rehearse with AI one question at a time with feedback, practise aloud, and do at least one mock with a real person.
Real-World Example: Hafiz asked a chatbot to interview him for an audit internship. For "Tell me about a time you found a mistake," he used STAR: as treasurer of his faculty club (S), he had to close the year's accounts (T); he matched every receipt to the bank statement and found RM640 recorded twice (A); the corrected accounts were approved at the AGM and he set up a monthly check (R).
Q: In a STAR answer, which part should usually take up most of your time?
The Action is where you show your own skills and judgement, so it deserves the most time. Keep the Situation brief, and say "I" rather than "we" so the interviewer hears your part.
Role-play: I will be the interviewer for the internship you want. Tell me the role and the company, and I will ask you one behavioural question. Answer using STAR, and after your answer I will give you feedback and one follow-up question.
Red Flags
Students looking for internships and part-time work are a favourite target of job scammers. Bukit Aman's Commercial Crime Investigation Department reported 1,537 job scam cases, with losses of RM31.8 million, in the first three months of 2026 alone. Learning the warning signs takes ten minutes and can save you a great deal.
Warning signs of a fake offer
Be suspicious when an "offer" arrives on WhatsApp or Telegram from an unknown number, with no official email or job portal involved. Be suspicious of pay that is too good for simple tasks, such as liking posts or rating products. Legitimate employers do not ask you to pay a registration, training, deposit or equipment fee before you start. Watch for pressure to decide quickly, an offer with no interview, and early requests for your bank account number, a copy of your IC or your passwords. Some scams pay a small commission first to win your trust, then ask you to "top up" larger amounts.
Never agree to let your bank account receive and pass on money for someone else. That makes you a
money mule, and mule accounts are investigated by the police even when the account holder was tricked.
Spot It, Then Check It - Job Scam Warning Signs and the Checks to Run
How to check an employer
Search the company on
SSM e-Search, the Companies Commission of Malaysia's registry check, to confirm it is registered. Contact the company through the phone number or email on its own website, not the one in the message. Check suspicious phone numbers and bank accounts on
Semak Mule, run by the Royal Malaysia Police. Your university LI unit can also tell you whether an employer is known to it. A real employer will need your IC and bank details for payroll, but only after a formal offer, through its HR department.
If you have already transferred money, call the
National Scam Response Centre (NSRC) on 997, which runs 24 hours a day, as soon as possible, then contact your bank and make a police report.
Red flags in real placements
Not every warning sign is a scam. If a genuine placement gives you no supervisor, only personal errands or unsafe tasks, raise it with your university supervisor early.
Watch video: Red Flags
Key Insight: Fake offers often come only through WhatsApp or Telegram, ask for a fee or bank details early, and promise high pay for simple tasks. Check the company on SSM e-Search and numbers on Semak Mule, and if you have lost money, call the NSRC on 997 at once.
Real-World Example: A week after posting her resume online, Aina received a WhatsApp message offering a "remote PR internship" at RM150 a day for liking clients' posts, with a RM50 registration fee. She searched the company name on SSM and found nothing, checked the number on Semak Mule, and forwarded the message to her university LI unit, which warned other students.
Q: A recruiter you have never spoken to messages you on Telegram with an internship offer, and asks for a RM100 training fee and your bank account number. What should you do?
An unsolicited offer on a messaging app, an upfront fee and an early request for bank details are three classic warning signs. Pay nothing, verify the company on SSM e-Search and the number on Semak Mule, and ask your LI unit.
Scenario: a friend in your course forwards you a message offering a "paid social media internship" at RM200 a day, applied for through a Telegram link. I will play your friend, who is excited and about to sign up. Talk me through what you would say and which checks we should run together.
Module 3: Before Day One
Arriving as the intern who already did the reading
Learn the industry's language, read the company like an insider, understand Malaysian workplace culture and communication, plan your first 30 days and start the internship log you will use for the rest of the course.
Learning Objectives - Build a verified, bilingual glossary of the terms your industry uses
- Prepare a one-page company brief from public sources, using AI only on public documents
- Apply Malaysian forms of address, feedback norms and religious and festive courtesies at work
- Write clear work emails, formal Malay openings and closings, chat replies and requests for help
- Draft a 30-day learning plan and start a daily internship log
What You'll Learn - A one-week glossary routine, checked against trusted sources
- Reading a company's website, news, annual report and house style
- Encik, Puan, Cik, Tuan, Dato', Datuk, Tan Sri and Dr
- Indirect feedback, face, prayer times, Ramadan and open houses
- Email structure and formal Bahasa Melayu conventions
- WhatsApp etiquette and asking for help: what I tried, what I need, by when
- A four-week plan: orient, practise, own, review
- The daily internship log: four questions, kept confidential
Learn the Industry's Language in a Week
Every workplace has its own vocabulary, and on your first day you will hear it everywhere. In a PR agency,
PR means public relations; in a software team, the same two letters mean a pull request, a proposed change to the code. An audit senior may ask for the
PBC list, the documents "prepared by client". A translation project manager may mention the
TM, or translation memory, and an editor may say a story is under
embargo, meaning it cannot be published before a set time. Interns who do not know these words spend their first fortnight nodding and guessing. Interns who learn them beforehand can follow meetings from day one.
A one-week glossary routine
You do not need a textbook. Spend about thirty minutes a day for a week. On days one and two, collect terms: read the company website, the job advertisement, two or three recent industry news articles and any case studies the company publishes, and note every word or abbreviation you could not explain to a friend. Aim for 30 to 50 terms. On days three and four, define them. On day five, write one example sentence for each, the way your new colleagues would use it. On days six and seven, test yourself and cut the list to the twenty terms you are most likely to hear.
Using AI, then checking it
An AI chatbot is a fast first explainer. Ask it to define a term for a new intern in your industry, give an example sentence, and say how the meaning changes in other fields. Then check each definition against a source the industry trusts, such as a professional body's glossary, a textbook or the company's own materials. AI can sound confident while mixing up meanings between industries or countries, and a glossary is only useful if it is right.
Make it bilingual
Many Malaysian offices switch between English and Malay in the same meeting, and documents for government or the public are often in Bahasa Melayu. Add the Malay equivalent next to each term where one exists, and check official Malay terms in
PRPM, the online language reference of Dewan Bahasa dan Pustaka. For languages and translation students, this glossary is also the first version of a term base, a habit your future employer will expect.
Key Insight: Collect 30 to 50 terms, define them, use each in a sentence, then keep the 20 you will hear most. Let AI explain first, but confirm every definition against a source your industry trusts.
Real-World Example: Aina's list for Teras Komunikasi started with boilerplate (the standard "About the company" paragraph at the end of a press release), media drop, embargo, coverage report, KOL and holding statement. Her AI chatbot defined KOL correctly as key opinion leader, but described a holding statement too loosely, so she checked it against a PR textbook from her faculty library and rewrote the definition in her own words.
Q: Why does this section tell you to check AI definitions against an industry source?
AI is a quick first explainer, but it can blend meanings from different industries or countries while sounding certain. A glossary is only useful if it is right, so each definition is confirmed against a professional glossary, a textbook or the company's own materials.
Action step: open your future company's website and one recent news article about its industry. List ten terms you could not explain to a friend, then tell me the three that confuse you most and we will work through them together.
Read the Company Like an Insider
In Module 2 you researched the company to win the placement. Now the question is different. You are not trying to impress an interviewer; you are trying to understand how the place works so you can be useful quickly. Insiders know four things that outsiders usually do not: what the organisation sells and to whom, how it makes money, what it cares about this year, and how it sounds when it speaks.
Where to look
Start with the company's own words. Read the About page, the services or products pages, the news or press room and any case studies. Scroll back through six months of its social media posts to see what it celebrates and how it writes. If the company or its parent is listed on
Bursa Malaysia, its annual report is public and explains the business, its risks and its plans. For a small private firm there may be little written down, so read about its clients and its industry instead.
As you read, build a one-page company brief: what it does, its main clients or customers, three recent news items, its competitors, and how it writes. That last point matters most for communication students. Many organisations have a
house style, a set of rules for spelling, tone, dates, titles and formatting. If you spot it early, your first drafts will need fewer corrections.
AI as a reading partner
AI tools are good at helping you digest long public documents. You can load an annual report into a tool such as Gemini Notebook (formerly NotebookLM) and ask what the main business segments are, or which risks the company names. Keep two limits in mind. Use only public documents at this stage, never anything the company has sent you privately. And treat the answers as leads to check in the original, not as facts.
The unwritten rules
Some things are not written anywhere: whether people reply on email or chat, how early meetings really start, what people wear on Fridays. Career adviser Gorick Ng calls these "the unspoken rules" in his book of that name. You learn them by watching closely in your first week, and by asking a friendly colleague rather than guessing.
Watch video: Read the Company Like an Insider
Key Insight: Insiders know what the organisation sells and to whom, how it makes money, what it cares about this year and how it sounds. Capture all four in a one-page company brief before day one.
Real-World Example: Hafiz could not see his audit firm's client files before starting, and did not try. Instead he read the firm's website, its recent graduate recruitment posts and the published annual report of one listed company in an industry the firm serves. When his senior mentioned a "going concern" review and a "management letter" in week one, he had already met both terms.
Q: Before your internship starts, which material is suitable to load into an AI tool to help you understand the company?
At this stage only public documents go into an AI tool. Anything the company has sent you privately, such as an onboarding pack, contract or handbook, stays out, and even answers about public documents are leads to check in the original.
Role-play: I will play your supervisor on your first morning. I will ask, "So, what do you know about us so far?" Answer in two or three sentences using your company brief, and I will tell you what landed well and what an insider would add.
Malaysian Workplace Culture
Malaysian offices bring together people of different ethnicities, religions, languages and ages, and most run smoothly because people pay attention to small courtesies. You do not need to master every custom before day one. You need to notice, ask politely and follow the lead of your colleagues.
How to address people
In Malay,
Encik is the polite form for a man,
Puan for a married woman and widely used for professional women generally, and
Cik for an unmarried woman.
Tuan is a respectful form for a man, common in government and in formal letters, where "Tuan/Puan" is the standard greeting. Doctors keep
Dr. Honorary titles come from state rulers and governors or from the federal government:
Dato' is usually a state title and
Datuk a federal title or one awarded by a state with a governor, though spelling varies by state.
Tan Sri is a senior federal title. Wives are usually styled Datin and Puan Sri. Copy exactly how the person writes their own title in their email signature, and when unsure, ask a colleague. Many start-ups use first names, so follow what others do.
Feedback and face
Feedback in many Malaysian offices is indirect. "Maybe you can relook this part" or "Can consider another angle" often means the draft needs real changes. Saving face matters on both sides, so raise disagreements privately rather than in front of others, and thank people for corrections.
Faith, festivals and the working week
Muslim colleagues pray at set times through the day, and Friday prayers mean a longer Friday lunch break in many offices; government offices typically break from 12.15pm to 2.45pm on Fridays. Kedah, Kelantan and Terengganu have a Friday-Saturday weekend, while Johor returned to Saturday-Sunday in January 2025. During Ramadan, which falls about 11 days earlier each year, many colleagues are fasting, so avoid booking lunch meetings and expect some offices to shorten hours. Around Hari Raya, Chinese New Year, Deepavali, Christmas and other festivals, colleagues may hold
open houses. If you are invited, go, dress modestly and enjoy it. When arranging office food, check that it is halal and ask about vegetarian needs.
Key Insight: Copy the exact title a person uses, read indirect feedback as real feedback, and plan around prayer times, Ramadan and the weekend pattern of the state you work in.
Real-World Example: Aina's supervisor, Puan Rosnah, returned her first media list with one comment: "Maybe can relook the KOL section." Aina first read it as optional. A colleague explained that it meant the section needed redoing, so Aina rebuilt it that afternoon and thanked Puan Rosnah privately for the pointer.
Q: Your supervisor writes "Maybe you can relook this part" on your draft. Based on this section, what does it most likely mean?
Feedback in many Malaysian offices is indirect, so a soft phrase like "maybe you can relook" usually signals that real changes are needed. Treat it as a clear request, act on it promptly and thank the person privately.
Role-play: I will play Puan Rosnah, your supervisor at a PR agency. I have just told you, "Can consider a different opening for this press release, maybe." Reply to me as you would in the office, showing that you understood what I really meant.
Workplace Communication Basics
Most of your work as an intern will be judged first through messages: emails, chat replies and the way you ask for help. Getting the basics right makes you easy to work with long before anyone sees your best work.
The structure of a work email
A good work email has a specific subject line ("Draft media list for Kopi Kita launch, for review by Thursday" rather than "Hi"), a greeting, one sentence saying why you are writing, the details in short paragraphs, a clear request with a deadline, and a polite closing with your name and role. Keep to one topic per email, so it can be found and forwarded later. Before sending, read it once as the receiver: could they act on it without writing back to ask what you mean?
Formal Malay openings and closings
Formal emails and letters in Bahasa Melayu follow set patterns. A common greeting is "Tuan/Puan," and a person with an honorary title is addressed as "Yang Berbahagia", as in "Yang Berbahagia Dato' Ahmad". The opening line often reads "Dengan hormatnya perkara di atas adalah dirujuk." A frequent closing is "Sekian, terima kasih." followed by "Yang benar," before your name. Government letters have used "Saya yang menjalankan amanah" since 2018, replacing the older "Saya yang menurut perintah". Copy the patterns your office already uses.
WhatsApp and group chats
Many Malaysian teams coordinate on
WhatsApp or a workplace chat app. Reply to work messages during working hours, acknowledge instructions briefly ("Noted, will send by 3pm"), and keep personal chat and stickers out of work groups. Send files to the person who needs them rather than the whole group, and never forward client material outside it.
Asking for help well
Supervisors want you to ask, but they want good questions. Use three parts:
what I tried,
what I need, and
by when. That shows effort, makes the request concrete and lets the other person plan. Gather small questions and ask them together rather than one at a time. AI can help you draft any of these messages, but check that the tone sounds like you and suits the person you are writing to.
Watch video: Workplace Communication Basics
Key Insight: Specific subject line, one topic, a clear request with a deadline. In Malay, follow the set patterns your office uses. When you ask for help, say what you tried, what you need and by when.
Real-World Example: Darren was stuck on a subtitle file at Lingua Rakan. Instead of "Sorry, I am stuck", he wrote: "Hi Kak Farah, I am timing the subtitles for the Malay version. I tried the method in our style guide, but the lines still overlap after 12:40. Could you show me how you usually fix this? I need it ready for the client check at 4pm." She sent him a two-minute fix.
Q: Which message follows the three-part structure for asking for help taught in this section?
A good request for help says what you tried, what you need and by when. That shows you made an effort, makes the request concrete and lets the other person fit it into their day.
Practice: write me a short message asking a colleague for help with a real or imagined task, using what I tried, what I need and by when. I will play the colleague and tell you honestly whether I could act on it straight away.
Your 30-Day Learning Plan
The first month of an internship passes quickly, and interns who arrive without a plan often spend it waiting for instructions. A simple 30-day learning plan changes that. It tells you what to focus on each week, gives your supervisor something concrete to react to, and turns a vague hope to "learn a lot" into goals you can check.
Orient, Practise, Own, Review - A Four-Week Plan for a New Intern
Four weeks, four jobs
In
week one, orient. Learn names and roles, the tools and systems, where files live, who approves what, and the house style. Your aim is to stop asking where things are by Friday. In
week two, practise. Do small tasks carefully and ask to see good examples of past work, because copying a good model is the fastest way to learn a format. In
week three, own something. Ask to take responsibility for one recurring task, such as daily media monitoring, a weekly report or a set of test cases, so that someone relies on you. In
week four, review. Ask your supervisor for feedback, compare what you can do now with day one, and suggest one small improvement you have noticed.
One skill, two people, one output
For each week, write down one skill you want to build, two people you want to learn from or meet properly, and one piece of work you want to finish. Link your skills to what your university will assess; the LI assessment form tells you which qualities your supervisor will score.
Drafting it with AI
Give an AI tool the job advertisement, your glossary and your company brief, and ask for a draft four-week plan for an intern in that role. Then edit hard. Cut anything you cannot actually do, add what you know about the company, and make every goal specific enough to tick off. Finally, show the plan to your supervisor in your first week and ask, "Is this what you would want me to focus on?" A plan they have adjusted is worth far more than a perfect one they never saw.
Watch video: Your 30-Day Learning Plan
Key Insight: Week one orient, week two practise, week three own something, week four review. Each week, pick one skill, two people and one output, and show the plan to your supervisor in your first week.
Real-World Example: Mei Ling's plan for the fintech start-up set week one as "set up my development environment and read the team's coding guidelines", week two as "fix two small bugs through code review", week three as "own the test cases for one feature" and week four as "present what I learned at the team meeting". Her supervisor moved code review earlier and added a meeting with the compliance lead.
Q: According to this section, what is the main focus of week three in the 30-day plan?
Week three is about owning something: one recurring task, such as media monitoring or a weekly report, that the team comes to rely on you for. Week one orients, week two practises and week four reviews.
Role-play: I will be your supervisor at your first-week check-in. Share your draft 30-day plan in a few lines, and I will respond the way a busy supervisor might, pushing back on anything vague or unrealistic.
Start Your Internship Log
Your memory of a busy week fades faster than you think. By week eight you will struggle to recall what you did in week two, and that is exactly when you will need it: for weekly updates, your LI report and your resume. The answer is an
internship log, a short private record you write every working day. It takes about five minutes, and it is the habit that runs through the rest of this course.
The Internship Log - Four Daily Questions That Feed Weekly Updates, the LI Report and Resume Bullets
Four questions a day
Each entry answers four questions.
What I did: the tasks, with enough detail to describe them later, such as "drafted three social posts for the launch".
What I learned: a skill, a term or a lesson from a mistake.
What I am stuck on: anything blocking you.
One thing to ask: a question for your supervisor or a colleague. Add the date and, where possible, evidence such as a number, a file name or feedback you received.
Why it pays off
The log feeds three later tasks in this course. Your weekly update to your supervisor becomes a summary of five entries. Your LI report draws on months of dated evidence instead of guesswork. And your resume bullets come from real tasks and results. It is not the same as your university's official logbook, which has its own format, but it makes filling in that logbook much easier.
Keep it safe
Write your log somewhere private, such as a notes app or a document on your own account. Do not record client secrets, passwords or other people's personal details; write "client A" or "the launch project" instead. If you later use AI to summarise a week of entries, paste only entries you have already cleaned of anything confidential. Module 5 explains why this matters.
Start today
You do not have to wait for day one. Write your first entry now about your preparation: the glossary you built, what you learned about the company, what still worries you and one thing you want to ask on your first morning.
Key Insight: Five minutes a day, four questions: what I did, what I learned, what I am stuck on, one thing to ask. Add dates and evidence, and keep anything confidential out.
Real-World Example: One of Aina's early entries read: "Thu 4 Mar. Did: built a media list of 25 lifestyle editors for client A. Learned: our house style writes dates as 4 March 2027, not 4/3. Stuck: two editors have left their outlets and I do not know who replaced them. Ask: Puan Rosnah, do we keep a master contact list?" That entry later became a line in her LI report.
Q: Which of these is one of the four questions in a daily internship log entry?
The four questions are what I did, what I learned, what I am stuck on and one thing to ask. The log records your own learning and progress, never other people's mistakes or anything confidential.
Action step: write your first internship log entry now, about your preparation so far, using the four questions. Paste it here and I will suggest how to make each answer more specific without adding anything confidential.
Module 4: Doing Real Work with AI
Faster first drafts, without bluffing
How to turn a vague brief into clear work, research and summarise with sources you can defend, check everything before it leaves your hands, and use AI for writing, translation and everyday office tasks without losing the trust you were given.
Learning Objectives - Decode a brief into its audience, purpose, format, deadline and constraints before starting work
- Trace AI-assisted research back to primary sources and write summaries you can defend
- Apply the five-check Verify-Before-You-Send checklist to any piece of work
- Adapt AI drafts to an organisation's house style, including the anatomy of a press release
- Post-edit machine translation to the right level and handle everyday office tasks safely
What You'll Learn - The five questions inside every brief
- Why AI tools hallucinate, and how to trace a claim
- The Verify-Before-You-Send checklist
- House style and press release anatomy
- Register, idioms, honorifics and code-switching
- Light and full post-editing of machine translation
- Emails, meeting notes and spreadsheets with AI
- Keeping client data and proprietary code out of public tools
Decoding a Brief
A
brief is the instruction you are given before a piece of work: what to make, for whom, and by when. Some briefs are a tidy one-page document. Many, especially for interns, arrive as a single line in a WhatsApp message or a hurried sentence as your supervisor walks past your desk. The work you produce can only be as good as your understanding of that instruction, so the first skill in this module happens before you open any AI tool.
Five questions inside every brief
Every brief, however short, hides answers to five questions.
Who is it for? A journalist, a client, your manager or the public each need something different.
What is it for? To inform, persuade, record a decision or get approval.
What does finished look like? Format, length, language and file type.
By when? And is that the final deadline or the draft deadline?
What do I already have, and what must I avoid? Templates, past examples, approved facts, and anything that is off limits.
When a brief is vague, write down the answers you can work out and turn the gaps into questions. Ask them in one message rather than five, and offer your best guess so your supervisor only has to say yes or correct you. "I will do the list in Excel with name, outlet, beat and email, English and Malay media only, by Thursday noon. Is that right?" takes ten seconds to answer.
Where AI helps, and where it does not
AI is useful here as a questioning partner, not as a mind reader. Paste the brief (with confidential details removed) and ask: "What questions should I ask before I start this?" or "What would a good version of this usually include?" The tool can point out gaps you missed, such as the language of the outlets or the file format. What it cannot do is tell you what your supervisor actually meant. Only your supervisor can confirm that, which is why the final step is always to restate the brief back to the person who gave it.
Key Insight: Before you start, restate the brief in one message: who it is for, what it is for, what finished looks like, the deadline, and what you already have. Offer your best guess so it can be confirmed in seconds.
Real-World Example: Aina's supervisor at Teras Komunikasi messages: "Can you do a quick media list for the Bayu launch?" Aina asks an AI tool what a media list usually contains, then replies: "I will list 30 lifestyle and business journalists in English, Malay and Chinese outlets, with name, outlet, beat and email, in our Excel template, by Thursday noon. Should I include radio?" Her supervisor answers in one line, and Aina avoids a day of rework.
Q: According to this section, what is the best way to deal with a vague one-line brief?
Correct. Working out what you can, then restating the brief in one message with your best guesses, lets your supervisor confirm or correct you in seconds. AI can suggest questions to ask, but only the person who gave the brief can say what they meant.
Role-play: I will be your supervisor. I walk past your desk and say, "Can you put together something on our competitors for Friday?" and then I head into a meeting. Write the one message you would send me to decode this brief, and I will reply in character.
Research and Summaries You Can Defend
AI chat tools are fluent, fast and often right. They are also capable of producing a confident sentence, a neat statistic or a perfectly formatted reference that does not exist. This is called a
hallucination: the tool generates text that sounds plausible because it predicts likely words, not because it has checked a fact. It is not lying, and it does not know it is wrong, which is exactly why the output reads so convincingly.
The most famous warning came from a New York court in 2023. In
Mata v. Avianca, two lawyers filed a brief citing court cases that ChatGPT had invented. The judge fined them and their firm US$5,000 in total, and noted that using AI was not improper in itself; the failure was not checking. If trained lawyers can be caught out, an intern on a deadline certainly can.
Trace every claim to a primary source
A
primary source is the original place a fact comes from: the official statistics release, the company's annual report, the law itself, the study's own paper. For Malaysian figures, that often means the Department of Statistics, whose open data portal
OpenDOSM publishes the numbers directly. A news article quoting a figure is a secondary source; an AI answer with no link is not a source at all.
Use AI to find leads, explain unfamiliar terms and suggest where to look. Then open the source, find the exact sentence or table, and check that the number, the date and the meaning all match. If you cannot find it, the claim does not go into your work. If the figure matters, tell your supervisor you could not confirm it.
Summaries that stay honest
When you summarise a report, tools that answer only from documents you upload, such as Gemini Notebook, reduce the risk because they cite the passage they used. Still click each citation. Keep what the source says separate from your own interpretation, and add a short
source note at the end: what you used, with links and page numbers. A summary you can defend is one where your supervisor can check any line in under a minute.
Watch video: Research and Summaries You Can Defend
Key Insight: AI can find leads and explain terms, but it is not a source. Every fact in your work should point to a primary source you have opened and read yourself.
Real-World Example: Mei Ling is asked to summarise a regulator's discussion paper for her fintech team. She uploads the PDF to a document-grounded tool, asks for the five main proposals with page references, then checks each one against the page. One "proposal" turns out to be a question the paper raised, not a decision. She corrects it and adds a source note listing the paper's title, date and the pages she used.
Q: Why does an AI hallucination often read so convincingly?
Correct. A chat tool generates likely text rather than verifying facts, so an invented statistic or reference can be just as fluent as a real one. That is why the fix is to trace every claim to a primary source yourself.
Action step: take one AI answer you have received recently that contained a fact or figure. Find the primary source for it and tell me what you found. Did the source say exactly the same thing, something slightly different, or nothing at all?
The Verify-Before-You-Send Checklist
Everything in this course comes back to one habit: nothing leaves your hands until you have checked it. The
Verify-Before-You-Send checklist turns that habit into five quick questions. You will see it named again in later modules, so it is worth learning by heart now.
The Verify-Before-You-Send Checklist - Five Checks Before Any Work Leaves Your Hands
The five checks
Facts and figures traced to a source. Every number, date and factual claim can be pointed back to a primary source you have opened yourself, as in the previous section.
Names, titles and quotes correct. A misspelt client name or a wrong title such as "Datuk" for someone who is "Dato'" is the error people remember. Quotes must be real words from a real person, approved for use.
Nothing confidential went into a public tool. Check what you pasted along the way, not just the final document. Client names, unreleased figures, personal data and code belong only in tools your organisation has approved.
It sounds like you and follows the house style. Remove the generic AI phrasing, and match the organisation's spelling, format and terminology.
You can explain every line if asked. If your supervisor points to any sentence and asks "Why did you say this?", you have an answer. If you do not, check it or cut it.
Notice that the checklist does not ask whether you used AI. It asks whether the work is accurate, safe, in the right voice and truly understood by you. Those are the questions a supervisor cares about, whichever tools helped you get there, and they apply just as much to work you wrote entirely by hand.
Making it a habit
The checklist takes two or three minutes for an email and longer for a report. Keep a copy at the top of your notes or on a sticky note by your screen, and run through it every time, including when you are rushing. Rushed work is when mistakes slip through, and it is also when a supervisor is least forgiving of them.
Real-World Example: Before sending a draft social media post to her supervisor, Aina runs the checklist. Check 1: the "40% growth" figure is in the client's approved fact sheet. Check 2: she spots that the AI draft wrote the CEO's name without the "Dr" she uses. Check 3: she pasted only the public product description into the tool. Check 4: she changes "elevate your lifestyle" to plainer words the client uses. Check 5: she can explain every line. Two fixes, three minutes.
Q: Which of these is one of the five checks on the Verify-Before-You-Send checklist?
Correct. "You can explain every line if asked" is the fifth check. The other four are facts traced to a source, names, titles and quotes correct, nothing confidential in a public tool, and sounding like you while following the house style.
Which of the five checks do you think you are most likely to skip when you are rushing, and why? Tell me, and we will work out one small trigger that would remind you to do it every time.
Writing in the Organisation's Voice
Every organisation has a voice, and many write it down in a
house style guide: the rules for spelling (British or American), dates, numbers, job titles, the company name and the words it avoids. A bank sounds careful. A start-up sounds casual. A government agency writes in formal Malay. AI drafts, left alone, sound like none of them; they sound like AI, with the same tidy, slightly grand phrases everywhere.
Teach the tool the voice
The fastest way to fix this is to show, not tell. Give the tool two or three approved past pieces (public ones, or ones your organisation allows you to use), plus the key house style rules, and ask it to draft in that voice. Then edit by hand. Read your draft aloud: if a sentence is one you would never say, rewrite it. Check the result against the style guide line by line, because tools drift back to their own habits.
The anatomy of a press release
For communication students, the press release is the classic test of voice and structure. Journalists receive many every day, so the format is predictable on purpose.
Anatomy of a Press Release - Six Parts Journalists Expect to Find
The
headline states the news. The
dateline gives the city and date. The
lead paragraph answers who, what, when, where and why, because many readers stop there. The body adds detail and
quotes. The
boilerplate is the standard paragraph about the organisation, and the
media contact tells journalists whom to call.
Quotes are never invented
AI will happily write a quote and put it in your CEO's mouth. A drafted quote can be a starting suggestion, but it becomes a quote only when the named person has seen it, agreed to it and approved it, usually in writing. Publishing words a real person never said is a serious breach of trust, with them and with the journalists who print it.
Watch video: Writing in the Organisation's Voice
Key Insight: A quote in a press release must be the real, approved words of the named person. An AI-drafted quote is only a suggestion until that person has read and approved it.
Real-World Example: Aina drafts a launch press release for a Teras Komunikasi client. She gives the AI tool two of the agency's past releases and the client's style rules, then rewrites the draft by hand. The tool suggested a quote from the client's managing director. Aina marks it "DRAFT QUOTE - FOR APPROVAL" in yellow, and her supervisor sends it to the client, who changes half of it before signing off.
Q: In a press release, what is the boilerplate?
Correct. The boilerplate is the organisation's standard "About" paragraph, used in the same form on every release. The who-what-when-where-why paragraph is the lead, and the city and date make up the dateline.
Role-play: I will be a client's managing director. You have sent me a draft press release with a quote attributed to me that I have never seen before. I call you and ask, "Who wrote these words for me?" Respond as the intern, and I will reply in character.
Translation and Localisation
Machine translation is now good enough that it is part of routine work at many translation firms, which makes the human skills more valuable, not less.
Translation moves meaning from one language to another.
Localisation goes further and adapts the text to a market: currency, dates, examples, humour and what is polite.
What machines still miss
The first is
register, the level of formality. Official Malay documents use
bahasa baku, the standard formal language, and a casual machine rendering can sound wrong to a ministry reader. The second is
idioms: "kacang lupakan kulit" translated word for word becomes nonsense about peanuts. The third is
honorifics: titles such as Dato', Datuk, Puan and Tuan carry meaning that a machine may drop, change or mistranslate. The fourth is
code-switching, the everyday mixing of languages in Malaysia, from
bahasa rojak (mixed Malay and English) to Manglish. A tool may "correct" it into formal English or fail to understand it at all, even when the client wanted that casual tone in an advertisement.
For Malay terms, check the reference site of Dewan Bahasa dan Pustaka, PRPM (Pusat Rujukan Persuratan Melayu), which covers spelling, dictionary definitions and official terminology.
Post-editing: light or full
Post-editing means correcting machine translation output. The industry distinguishes two levels.
Light post-editing fixes only what stops the text from being understood or makes it wrong, and is used for internal or short-lived content.
Full post-editing aims for a result comparable to a human translation, correct in meaning, grammar, terminology and style. The international standard for this work, ISO 18587, sets requirements for full post-editing. Always ask which level a job needs, because over-editing wastes time and under-editing damages trust.
Subtitling basics
Subtitles have their own constraints: limited space and limited reading time. The
Netflix English Timed Text Style Guide, for example, allows two lines of up to 42 characters each, and caps reading speed for adult programmes at 20 characters per second. Rules differ by client and language, so follow the guide you are given. A good subtitle often means shortening the original, which machines rarely do well.
Watch video: Translation and Localisation
Key Insight: Ask which level of post-editing a job needs before you start. Light post-editing makes text accurate and understandable; full post-editing makes it as good as a human translation.
Real-World Example: At Lingua Rakan, Darren receives a machine-translated English version of a hotel's Malay brochure for full post-editing. The tool rendered "Datuk Lim" as "Grandfather Lim", translated a Malay proverb word for word, and turned a playful slogan into stiff, literal English. He restores the title, checks the meaning of two Malay terms on PRPM, and asks his supervisor whether the slogan should be translated or rewritten for English-speaking guests.
Q: What is the difference between light and full post-editing?
Correct. Light post-editing fixes errors that block understanding or change meaning, and suits internal or short-lived content. Full post-editing aims for a result comparable to a human translation in meaning, terminology and style.
Think of a Malay, Chinese, Tamil or Manglish phrase you use often with friends. How would a machine probably translate it into English, and what would a good translator do instead? Share the phrase and your version.
Everyday Office Work
Not every internship task is a press release or a translation. Much of the day is emails, meeting notes, spreadsheets, slides and small fixes. AI can make each of these faster, and each comes with one trap to avoid.
Emails, notes and slides
For emails, use AI to tighten a draft you have written, or to suggest a polite way to decline or chase. Keep your own facts and your own voice. For meeting notes, a tool can turn your rough notes into a tidy list of decisions and action items; recording or transcribing a meeting, though, needs the permission of the people in it and must follow company rules. For slides, AI can suggest a structure, but the numbers and claims still go through the checklist.
Spreadsheets: understand, do not outsource
AI is excellent at explaining a formula you inherited or suggesting one for a task, such as a lookup across two sheets. Test any formula on a small sample where you already know the answer before you apply it to real data. Never ask a chat tool to calculate final figures for you and copy them across; you need to be able to show how every number was reached.
Hafiz at the audit firm
In audit, client information is confidential by default. The
Malaysian Institute of Accountants By-Laws on Professional Ethics, based on the international code for accountants, list confidentiality as a fundamental principle. Hafiz therefore never pastes a client's ledger, names or figures into a public AI tool. When he needs help with a formula, he builds a small dummy table with invented numbers, asks about that, and applies what he learns inside the firm's own files.
Mei Ling at the fintech start-up
A company's source code is often its most valuable asset. In 2023, Samsung restricted staff use of generative AI tools after engineers pasted internal source code into ChatGPT. Mei Ling never pastes proprietary code into a public tool. She uses only the coding assistant her company has approved, and for general questions she writes a tiny made-up example that shows the problem without revealing anything real.
Module 5 covers these rules in depth. For now, the third check on your list is the one to remember most in everyday work.
Real-World Example: Mei Ling is stuck on a date-formatting bug. Instead of pasting the payment service's code into a public chatbot, she writes a five-line example with invented variable names that reproduces the same error, asks about that, and then applies the fix in the company's approved editor. She mentions in her internship log what she learned, so she can explain it if her lead asks.
Q: Hafiz needs help with a spreadsheet formula for a client's figures. What does this section say he should do?
Correct. A dummy table with invented numbers lets Hafiz learn the formula without exposing any client information. Deleting chat history afterwards does not undo the disclosure, and removing only the name still leaves confidential figures in a public tool.
Role-play: I am your team lead at a tech company. I see you have a public AI chatbot open next to our codebase, and I ask, "What exactly have you been putting into that?" Answer me honestly, and explain how you use AI without exposing our code.
Module 5: Rules, Data and Trust
The module that keeps you employed
What must never go into a public AI tool, how the PDPA and confidentiality apply to you, how to follow company AI policies, the copyright basics behind AI output, photos and translations, and when to say you used AI.
Learning Objectives - Classify work information as red, amber or green before it goes near a public AI tool
- Explain the PDPA's seven principles and the 2024 amendments as they apply to an intern
- Locate and follow your company's and university's AI rules, using the national AIGE guidelines as a reference
- Apply copyright basics to AI output, photos, translations and the work you create
- Check public statements for accuracy and decide when and how to disclose AI use
What You'll Learn - What public AI tools may do with your inputs
- Red, amber and green information
- PDPA 2010 and its seven principles
- The 2024 PDPA amendments
- Company AI policies and the AIGE guidelines
- Copyright, ownership and AI-generated works
- Accuracy, 3R sensitivities and deepfakes
- Disclosing your use of AI
What Never Goes into a Public AI Tool
Whatever you type into an AI chatbot goes to a company's servers. That is fine for a grammar question, not for a client's unreleased results. The rule is easy to remember: if you would not pin it on a public noticeboard, do not paste it into a public AI tool.
What "public" means here
In this course a
public AI tool is the free or personal version of a chatbot that you sign into with your own email. Your company may provide a work version instead, such as an enterprise account run by its IT team, which usually comes with a contract saying your inputs will not be used to train the model. Unless your supervisor has told you a tool is approved for work data, treat it as public.
What can happen to what you type
Settings differ and change, so read each tool's current privacy page. Broadly, several free tools may use your conversations to improve their models unless you switch that off. OpenAI, for example, lets you turn off a setting called "Improve the model for everyone" and offers Temporary Chats, which it says are not used for training and are kept for no more than 30 days. Google says some Gemini conversations are read by human reviewers, and that reviewed chats can be kept for up to three years even after you delete your activity. This does not mean your words will appear in a stranger's answer. It means you have lost control of where they go, which alone can break a confidentiality promise.
The risk is not theoretical. In 2023 Samsung restricted staff use of generative AI after engineers uploaded sensitive internal source code to ChatGPT.
The Traffic-Light Test - Sorting Information Before It Goes into a Public AI Tool
Red, amber, green
Red never goes in: other people's personal data such as names with IC numbers, phone numbers, health details or salaries; client documents; unreleased financial results; passwords and source code; and anything covered by an embargo or an
NDA (non-disclosure agreement, a contract promising to keep information secret).
Amber needs a pause: internal material with names and identifying details removed, and only if your supervisor agrees.
Green is fine: published information, your own general questions and made-up examples.
When in doubt, strip it down. Replace "Encik Rahman of Syarikat ABC" with "Client A", remove the figures, and ask about the structure rather than the substance. This is check 3 of your Verify-Before-You-Send checklist: nothing confidential went into a public tool.
Watch video: What Never Goes into a Public AI Tool
Key Insight: Treat any AI tool your organisation has not approved as a public noticeboard. If it could not go on the noticeboard, it cannot go in the chat.
Real-World Example: Aina is asked to tighten a press release about a client's merger, embargoed until Thursday. Instead of pasting it into her personal chatbot, she writes a short dummy version about "Company X" with invented figures, asks the AI how to make the lead paragraph stronger, then applies that structure herself to the real draft in the agency's own document. The AI helped, and the merger never left the building.
Q: Which of these belongs in the red category for a public AI tool?
Unreleased financial results are confidential client information, so they are red and never go into a public tool. Published news, invented examples and general questions are green because nothing private is exposed.
Action step: list five kinds of information you expect to handle during your internship, and sort each into red, amber or green. Share your list with me and I will challenge any item you have placed too generously.
PDPA and Confidentiality for Interns
Malaysia's main privacy law is the
Personal Data Protection Act 2010, or
PDPA. It protects
personal data, information about an identifiable living person, when processed in
commercial transactions. A journalist's mobile number on Aina's media list and a customer's IC number in Hafiz's audit file both count. The Act does not apply to the federal and state governments. This module is general guidance, not legal advice; for a real question, ask your supervisor or data protection officer.
Some personal data is
sensitive and needs stronger safeguards, usually explicit consent: physical or mental health, political opinions, religious beliefs, the commission of offences and, since 2025, biometric data such as fingerprints and face scans.
The seven principles
Organisations must follow seven principles.
General: process personal data only with consent or another lawful basis.
Notice and Choice: tell people what is collected and why.
Disclosure: do not share it for other purposes without consent.
Security: protect it from loss and misuse.
Retention: keep it no longer than its purpose needs.
Data Integrity: keep it accurate and up to date.
Access: let people see and correct their data.
What the 2024 amendments changed
PDPA Amendments - What Came into Force and When
The Personal Data Protection (Amendment) Act 2024 came into force in three stages during 2025. From 1 April 2025 the organisation is called a
data controller instead of a data user, service providers processing data for it must meet the Security Principle themselves, and the maximum penalty for breaching the principles rose to a RM1 million fine, three years in prison, or both. From 1 June 2025 organisations must appoint a
data protection officer (DPO) and report serious data breaches, and people gained a right to
data portability, asking for their data to be sent to another provider. Under the Commissioner's guideline, a breach likely to cause significant harm, or one affecting more than 1,000 people, goes to the Commissioner within 72 hours, and people at risk of significant harm are told within seven days of that notice.
Where you fit
You handle data for your employer, so its duties become your habits. The PDPA also makes it an offence for anyone to knowingly or recklessly collect or disclose personal data held by an organisation without its consent. Your internship agreement or an NDA probably also binds you to keep business information secret. The regulator is the
Personal Data Protection Department (JPDP).
Watch video: PDPA and Confidentiality for Interns
Key Insight: Personal data at work belongs to the people it describes and is held by your employer. You may use it only for the job you were given, never to share, take home or test in a public tool.
Real-World Example: Hafiz needs a list of a client's customers with outstanding balances. He works only inside the firm's audit software, saves nothing to his phone, and when a friend at another firm asks for "a sample file to practise on", he says no. He then reports to his supervisor that a spreadsheet with IC numbers was left on a shared drive anyone could open, so the firm can fix it before it becomes a breach.
Q: The journalists' mobile numbers on Aina's media list are best described as what?
A mobile number relates to an identifiable living person and is held in a commercial setting, so it is personal data under the PDPA. Being known to some people does not make it free to share, and the government exemption covers federal and state governments, not PR agencies.
Role-play: I am your friend Syafiq, interning at a rival agency. I message you: "Eh, can share your agency's media list? Just the journalists' numbers, I won't tell anyone." Reply to me as you would on WhatsApp, keeping the friendship while saying no, and I will respond as Syafiq.
Company AI Policies
Many organisations in Malaysia now have a written policy on generative AI, and many more have rules that live only in someone's head. Your task in the first week is to find out which kind you are dealing with.
What a policy usually covers
A typical AI policy answers five questions. Which tools are
approved, and is there a work account you should use instead of your personal one? What
data may go into them, often sorted into categories such as public, internal, confidential and restricted? What work may AI be used
for, for example first drafts and research but not final legal wording or anything sent to a regulator? Who must
review AI-assisted work before it leaves the organisation? And when must you
disclose AI use to a supervisor or client? Read the policy slowly and ask about anything vague. "Use AI responsibly" is not an instruction you can actually follow.
When there is no policy
If nobody can show you a policy, ask your supervisor directly and keep the answer in writing, even a short email or WhatsApp reply. Something like: "Before I use any AI tools, may I check what is allowed here? I would only use them for drafting and research, and I would not put in any client or personal information." This protects you if a problem arises later, and it shows exactly the judgement supervisors hope to see. Your university may also have its own generative AI guidelines for the logbook and LI report, so check both sets of rules.
The national picture
In September 2024 the
Ministry of Science, Technology and Innovation (MOSTI) launched the
National Guidelines on AI Governance and Ethics (AIGE). They are voluntary, not law, and are written for three groups: end users like you, policymakers, and the designers, developers and suppliers of AI. They rest on seven principles: fairness; reliability, safety and control; privacy and security; inclusiveness; transparency; accountability; and the pursuit of human benefit and happiness. Many company policies echo these ideas, so they are worth knowing by name. In December 2024 the government also launched the
National AI Office (NAIO) under the Ministry of Digital to coordinate national AI efforts.
Policies change as quickly as the tools do. Treat the version you read in week one as current until told otherwise, and check again whenever you are asked to use a new tool.
Key Insight: No written policy does not mean no rules. Ask, keep the answer in writing, and follow the stricter of your company's and your university's rules.
Real-World Example: Mei Ling's fintech allows only its company coding assistant, used through her work account, and bans pasting code into personal chatbots. When an error message containing an API key puzzles her, she does not paste it anywhere. She removes the key and the code, asks her supervisor whether she may paste the generic error text into a public tool, and gets a one-line "yes, that's fine" on Slack that she keeps.
Q: According to this section, what should you do if your company has no written AI policy?
A short written answer from your supervisor shows good judgement and protects you if questions come up later. Silence is not permission, and another company's policy does not reflect your organisation's clients, contracts or risks.
Role-play: I am Puan Rozita, your supervisor, on your second day. You have five minutes with me before a meeting. Ask me what the rules are for using AI tools here, and be specific about what you would like to use them for. I will answer as a busy but reasonable supervisor.
Copyright, Attribution and AI-Generated Content
Copyright is the legal right to control the copying of creative work, and interns create and reuse a lot of it. In Malaysia it is governed by the
Copyright Act 1987. As before, this is general guidance, not legal advice.
The basics
Copyright arises automatically once an original work is written down, recorded or otherwise put into material form. You do not need to register it, although
MyIPO, the Intellectual Property Corporation of Malaysia, offers voluntary notification as evidence of ownership. For literary, musical and artistic works, protection generally lasts for the author's life plus 50 years. The Act allows
fair dealing for purposes such as research, private study, criticism, review and news reporting, usually with acknowledgement of the title and author. Fair dealing is narrow, so "I credited it" is not a licence to copy.
Who owns what you make at work
Under
section 26, copyright first belongs to the author. However, where a work is made in the course of the author's employment, or is commissioned by someone who is not the author's employer, copyright is deemed transferred to the employer or the person who commissioned it, unless an agreement says otherwise. Interns are not always employees, so read your internship agreement, which may assign your work to the company. Either way, ask before adding work you drafted to your portfolio.
AI-generated content
Whether purely AI-generated work is protected in Malaysia is
unsettled. The Act has no provision for it, and much of it assumes a human author, for example by tying protection to the author's lifetime. In July 2026 MyIPO issued a public consultation paper asking whether AI-generated works should be protected, who should own AI-assisted works and how much human contribution should be required. Until the law is clearer, assume your own selection, editing and rewriting is what makes AI-assisted work yours, and keep a record of what you changed.
Photos, stock and translations
A photo found on Google is not free to use. Press photos belong to the photographer or agency, and stock images come with licences that limit where and how they may appear, so check the terms and credit line. Translations are
derivative works: section 8 protects a translation as an original work, but without prejudice to copyright in the source, so someone needs the right to translate the original in the first place.
Key Insight: Assume every image, text and recording belongs to someone. Find out who, check what the licence allows, and credit as required. Crediting alone is not permission.
Real-World Example: Darren is asked to translate a Chinese product brochure into Malay for a client of Lingua Rakan. Before starting, he checks with his project manager that the client owns or has licensed the original text, and points out that two photos in the brochure carry a stock agency watermark, so the client must confirm the licence covers a Malay edition. When he uses machine translation for the first draft, he notes in his file which passages he rewrote.
Q: Under section 26 of the Copyright Act 1987, who owns a work made in the course of the author's employment?
Copyright first vests in the author, but section 26 deems it transferred to the employer for works made in the course of employment, subject to any agreement excluding or limiting that transfer. Registration plays no part, because Malaysian copyright arises automatically.
Opinion check: should a poster, poem or press release produced entirely by an AI tool, with no human editing, receive copyright protection in Malaysia? Give me your view and one reason, and I will argue the other side so you can test your thinking.
Accuracy and Public Statements
Anything published, whether a social post, a media statement or a reply to a customer on Facebook, speaks for the whole organisation. An internal error costs a correction. A public one can cost a client, a regulator's attention or a trending hashtag.
Why AI raises the stakes
AI tools write fluent, confident text that can contain invented statistics, misattributed quotes, out-of-date facts and sources that do not exist. The more polished the draft, the easier it is to wave it through. For public material, trace every figure to a source you can name, confirm every quote with the person who said it, and send the piece through your organisation's approval chain, however small the post. Interns should not publish anything without sign-off.
Sensitive topics
In Malaysia, content touching on race, religion and royalty, often called the
3R, is handled with particular care, both by the law and by audiences. This does not mean avoiding these subjects altogether. It means recognising that a joke, a careless translation or an AI-generated image can be read very differently across communities. If a draft touches any of the 3R, flag it to your supervisor before it goes anywhere.
The law on online content
Section 233 of the
Communications and Multimedia Act 1998 covers improper use of network services, such as knowingly sending content that is obscene, indecent, false, menacing or grossly offensive with intent to annoy, abuse, threaten or harass another person. Amendments in force from 11 February 2025 replaced the word "offensive" with "grossly offensive", added content intended to commit fraud or dishonesty, and raised the maximum penalty to a RM500,000 fine, two years in prison, or both. It is a reminder that posting is never consequence-free.
Deepfakes
A
deepfake is AI-generated audio, video or imagery made to look like a real person. In 2024 an employee at the Hong Kong office of engineering firm Arup transferred about US$25 million after a video call in which the "chief financial officer" and colleagues were deepfakes. In Malaysia, The Star has reported deepfake advertisements using the faces of the King and the Prime Minister to promote non-existent investments. Never act on an urgent request for money or data from a call or voice note alone; confirm through a channel you already know. Before sharing a viral claim, check
Sebenarnya.my, the fact-check portal run by the
MCMC.
Watch video: Accuracy and Public Statements
Key Insight: Public words carry the organisation's name, not yours. Every figure traced, every quote confirmed, every post approved, and every urgent request checked through a channel you already trust.
Real-World Example: An AI tool gives Aina a polished draft statement for a client's food bank launch, including the line "helping 40% of B40 families in the Klang Valley". She cannot find any source for the figure, so she deletes it, replaces it with the number of families the client has actually registered, confirmed by email, and marks the draft "for approval" rather than scheduling it herself.
Q: A video call from your "director" urgently asks you to transfer company funds. What does this section advise?
Deepfakes can copy a face and voice convincingly, as the Arup case showed, so the call itself cannot be trusted. Confirming through a separate, known channel, such as the director's usual phone number or your supervisor, is the safe step before any money or data moves.
Role-play: I am Encik Faizal, the account director. You have drafted a social media post for our client using AI, and I am reviewing it before it goes live. Tell me how you checked the facts, figures and any sensitive wording, and I will question anything that sounds unchecked.
When to Say You Used AI
Using AI is not dishonest. Hiding it when it matters is. The difficulty is that "when it matters" depends on who is asking and why, so it helps to have a clear rule of thumb you can apply in seconds.
Three questions to ask
First, does a
rule require disclosure? Your company policy, a client contract or your university's guidelines for the LI report may say exactly when and how to declare AI use. If so, follow the rule. Second, would the other person
expect to know? A supervisor who asks for a reflection on what you learned expects your own thinking, and a client who pays for human translation expects a human translator. If they would feel misled on finding out later, tell them up front. Third, how high are the
stakes? Grammar help on an internal email needs no announcement. A research summary that will shape a client recommendation does, because the reviewer needs to know which parts to check most closely.
How to say it
Disclosure works best when it is specific and brief. "I used ChatGPT to suggest a structure and tighten the wording; the research, figures and recommendations are mine, and I have checked every source" tells your reader what they need. "AI was used" tells them almost nothing. Name the tool, what it did and what you did. This fits the
transparency principle in the national AIGE guidelines, which call for openness about how AI is used rather than hiding it.
What disclosure does not do
Saying you used AI does not move responsibility onto the tool. If an AI-assisted report contains a wrong figure, it is still your wrong figure. That is why the fifth Verify-Before-You-Send check asks whether you can explain every line if asked. If you could not defend a sentence in a meeting, it should not carry your name.
Assessed work
Take most care with anything your university grades, such as the logbook, the LI report and the final presentation. Universities treat presenting AI-generated work as your own as a matter of academic integrity, and the rules differ between faculties. Read your own faculty's guidance before you start writing, not the night before submission, and keep brief notes on how you used AI as you go, so you can declare it accurately and confidently.
Key Insight: Disclose when a rule requires it, when the other person would expect to know, or when the stakes are high. Say which tool, what it did and what you did, and remember the responsibility stays with you.
Real-World Example: Darren's client contract at Lingua Rakan specifies "human translation". He used machine translation for a rough first draft before rewriting it fully. Rather than deliver quietly, he tells his project manager exactly what he did, and she checks with the client, who agrees in writing that a fully post-edited translation is acceptable for this job.
Q: Which of these is the most useful way to disclose AI use to a supervisor?
A specific disclosure tells the reader which parts came from the tool and which from you, so they know where to check. A vague label gives them nothing to act on, and silence can feel like concealment if it comes out later.
Role-play: I am Ms Tan, interviewing you for a graduate role six months from now. I ask: "Tell me how you used AI during your internship, and how you made sure the work was still yours." Answer as you would in the interview, and I will follow up with one harder question.
Module 6: Being Seen and Ending Well
Converting the internship into what comes next
How to make good work visible without bragging, take feedback well, build real relationships, write your LI report, and leave with a portfolio, a reference and possibly a return offer.
Learning Objectives - Write a short weekly update from your internship log that your supervisor can act on
- Ask for, receive and act on feedback, and recognise problems that belong with HR and your university, not AI
- Build professional relationships during a placement without forcing them
- Plan an LI report and final presentation from your own records, within your university's AI rules
- Turn internship work into honest resume bullets and portfolio pieces, and ask well for a reference or return offer
What You'll Learn - Weekly updates: Done, Next, Need from you
- Asking for specific feedback and decoding indirect feedback
- Difficult conversations and when to go to HR
- Networking inside the office and on LinkedIn
- The LI report, final presentation and AI disclosure
- Achievement-based resume bullets and the X-Y-Z formula
- Portfolio pieces, permission and confidentiality
- References, handover notes and return offers
Weekly Updates That Get Noticed
Plenty of interns do good work that nobody notices. It is not that supervisors are unkind. They are busy, they manage several people, and by the end of a month they genuinely cannot remember which intern fixed the media list or caught the error in the spreadsheet. The simplest fix is a
weekly update: a short message, sent at the same time every week, that tells your supervisor what you did, what comes next and what you need.
In your first week, ask your industry supervisor whether they would like one and in what form. Some prefer a Friday email, some a message in the team chat, some a five-minute conversation. Asking shows initiative, and their answer tells you how they like to communicate.
Three headings are enough
A useful update has three parts.
Done lists two or three things you finished this week, each with its result, not just the activity.
Next says what you plan to work on.
Need from you names any question, decision or blocker your supervisor can act on. That last part is what makes the update useful to them rather than just a report about you. Keep the whole thing under 150 words, so it can be read in a minute.
From Internship Log to Resume - How One Daily Habit Feeds Everything Else
Built from your log, not your memory
If you keep the internship log from Module 3, the update almost writes itself. On Friday, read that week's five entries and pick out the results, the questions and the things you are stuck on. An AI tool can help here: paste in your entries, with client names and anything confidential removed first, and ask for a draft under the three headings. Then edit it. Run it through the Verify-Before-You-Send checklist, paying attention to check 3 (nothing confidential went into a public tool) and check 4 (it sounds like you). Watch for inflated wording. AI loves verbs like "spearheaded", and your supervisor knows you formatted a spreadsheet.
The same log will later feed your LI report and your resume, so each weekly update is also a small rehearsal for the documents you will write at the end. You can practise now: write an update for a group project this week.
Key Insight: A good weekly update has three parts: Done, Next and Need from you. Keep it under 150 words, send it at the same time every week, and build it from your log rather than from memory.
Real-World Example: Aina's week 3 update at Teras Komunikasi read: "Done: two first drafts of the property client's press release, both approved after one round of edits; a 42-contact media list for the launch, checked against last month's coverage. Next: tracking coverage after Tuesday's launch. Need from you: should I follow up with journalists by phone or by email?" It took her ten minutes, because her log already held the details.
Q: Which part of a weekly update gives your supervisor something they can act on?
The Need from you heading turns the update from a report into a request your supervisor can answer. Done and Next matter too, but they inform rather than prompt action.
Action step: take one week of a university project, club role or part-time job and write a practice weekly update under three headings: Done, Next and Need from you. Paste it here and I will read it as your supervisor would on a busy Friday afternoon.
Feedback and Difficult Conversations
Feedback is the fastest way to learn during an internship, but "Any feedback?" rarely gets a useful answer. Busy people say "All good" and move on. Ask about one piece of work, soon after it is done, with a question that is easy to answer: "On the press release, what is one thing I should do differently next time?" Ask again at the mid-point of your placement for a fuller conversation about how you are doing overall.
Hearing what is actually meant
As Module 3 explained, feedback in many Malaysian offices is indirect, to protect face. "Maybe can try another way" may mean "this does not work". Check politely without putting your supervisor on the spot: "Just to make sure I understand, would you like me to rewrite the opening paragraph?" Then listen, thank them, and resist defending the work in the moment. Write the feedback in your log, apply it, and mention it in your next weekly update. Supervisors notice the intern who changes after being told once.
Conversations you would rather avoid
Some conversations are uncomfortable: you have too much work, instructions are unclear, or you made a mistake. Use a simple shape: what happened, what you have already done, and what you propose. Own mistakes early. A small error reported on Monday is a minor issue, and the same error found by a client on Friday is a serious one. AI is useful for preparing. You can rehearse the conversation with an AI tutor playing your supervisor, or turn vague feedback into a list of actions. Leave out colleagues' names and private messages when you do.
When AI is not the answer
Harassment of any kind, bullying, being asked to do something unsafe or illegal, or anything that makes you feel unsafe is not a writing problem, and an AI chat is not where it should go. Take it to people with responsibility for you:
HR at the company and your university supervisor, who oversees your placement and must know. Keep a dated, factual record of what happened. In Malaysia, the
Tribunal for Anti-Sexual Harassment, set up under the Anti-Sexual Harassment Act 2022, hears complaints from any person, not only employees, and Talian Kasih (15999) offers support. This is general information, not legal advice. In an emergency, call 999. Before you start, save your university's LI unit contact in your phone.
Watch video: Feedback and Difficult Conversations
Key Insight: Ask for specific feedback on specific work, check what indirect feedback means, and show you acted on it. Harassment, bullying and unsafe situations go to HR and your university supervisor, not to an AI tool.
Real-World Example: Darren's supervisor at Lingua Rakan said his subtitles were "a bit long". Rather than guessing, he asked: "Is it the length of each line or the time each subtitle stays on screen you would like me to cut?" It was the line length in the client's style guide. He logged it, fixed the next file, and his supervisor did not have to mention it again.
Q: During your placement, a senior colleague keeps sending you uncomfortable personal messages after work. What is the right first step?
Harassment needs people with responsibility for you, which means HR and your university supervisor, plus a factual record. AI can help you rehearse ordinary difficult conversations, but it cannot protect you or act on a complaint.
Role-play: I will be your industry supervisor. I have just looked at your draft and said, "Okay lah, but maybe can try a different approach for the opening." Reply as you would in the office, and I will tell you how your response landed and what I really meant.
Networking Without Being Awkward
The word "networking" makes many students picture name cards and forced small talk at a hotel ballroom. During an internship it is much simpler: getting to know the people around you, and letting them see how you work. Your network is already sitting in the office with you. It includes your supervisor, your team, the other interns, people in other departments, and the clients and suppliers you meet along the way. Many later opportunities, from a reference to a job lead, come from someone who watched you work and remembered you.
Inside the office
Start with the basics. Learn names and how people like to be addressed, greet people in the morning, and join team lunches when you are invited. Ask people about their work rather than about yourself: "How did you get into this role?" or "What does a busy week look like for you?" Most people enjoy explaining what they do to someone who is genuinely curious.
Once you have settled in, ask your supervisor whether you could have a short chat with someone in another team, perhaps 15 minutes over coffee. Prepare two or three questions, keep to the time, and thank them afterwards with one specific thing you learned. Be helpful in small ways too, especially to the other interns. The intern who shares a useful template is remembered warmly.
On LinkedIn
Connect on LinkedIn after you have actually met someone, and add a short personal note: "Thank you for walking me through how the team monitors coverage on Tuesday." Free accounts can add a note to only a few requests each month, so save notes for people you have met. Check your company's practice before connecting with clients. Before you post about your internship, check the social media policy and confidentiality rules, and never post client work before it is public or without permission.
Where AI helps, and where it shows
AI can help you prepare for a coffee chat by summarising someone's public role or suggesting questions, and it can draft a connection note. Write the final version yourself. A note that reads like every other AI-written message tells the reader you did not bother. You can prepare now by practising a 20-second introduction: who you are, what you study and what you are curious about.
Watch video: Networking Without Being Awkward
Key Insight: Networking during an internship means getting to know the people already around you. Ask about their work, keep coffee chats short and prepared, and connect on LinkedIn only after you have met, with a note that mentions something real.
Real-World Example: Mei Ling asked her supervisor at the Cyberjaya fintech start-up to introduce her to someone in the compliance team. She spent 15 minutes learning how they review new features before release. Three weeks later, when her own feature note needed a compliance check, she already knew whom to ask and what they would look for.
Q: You enjoyed a 15-minute chat with someone from the digital team. What is the best follow-up on LinkedIn?
A short note that mentions something real from the conversation reminds them who you are and shows you were listening. Posting details of client work risks confidentiality, and asking for a job straight away turns a conversation into a transaction.
Role-play: I am a senior manager from another department, and we have just met at the office pantry. Introduce yourself in about 20 seconds and ask me one question about my work. I will reply in character, then tell you what made your introduction memorable or forgettable.
Your LI Report and Final Presentation
Malaysian universities commonly ask for a written LI report at the end of industrial training, and many add a presentation or a short viva. These are often assessed alongside your logbook and your industry supervisor's evaluation. The required structure, length and marking scheme differ between universities and even between faculties, so the first rule is simple:
get your own faculty's template and rubric before you start, and follow it over anything you read here or anywhere else.
What reports often contain
Templates vary, but many ask for similar things: a background of the organisation and its structure, your role and the tasks you carried out, the skills and knowledge you gained, the challenges you faced and how you handled them, how the placement linked to your programme, and a conclusion with recommendations. Appendices often hold your logbook and selected work samples.
Assemble it, do not invent it
If you have kept your internship log and sent weekly updates, you already have the raw material: dates, tasks, results, problems and what you learned from each. Do not wait for the final week. Draft one section every few weeks while the details are fresh, and the report becomes an editing job rather than a last-minute reconstruction.
Confidentiality and AI
Your report will be read by lecturers and may be kept by the university. Ask your industry supervisor which details and samples you may include. Refer to clients in general terms, such as "a property developer", unless you have permission to name them, and remember that some companies ask to review the report before you submit it. Follow your university's policy on AI in assessed work, including any requirement to declare how you used it. If the policy is unclear, ask your university supervisor. Using AI to suggest a structure or check grammar is very different from asking it to write your reflections. The reflective parts must be yours, because they describe what only you experienced. And check 3 still applies: confidential material never goes into a public tool.
The presentation
Tell a story rather than reading out a task list: the problem, what you did, what changed and what you learned. Show evidence, such as an approved sample or a number from your log. Then rehearse with an AI tutor playing a panel that asks hard questions.
Key Insight: Follow your own faculty's template and rubric. Build the report from your log, drafting sections as you go. Get permission for samples, anonymise clients, follow your university's AI policy, and keep the reflections in your own words.
Real-World Example: For his "Challenges" section, Hafiz used his week 4 log entries about struggling with a bank reconciliation for an audit client. He wrote about "a manufacturing client" rather than naming it, and his supervisor approved two sample worksheets for the appendix once the figures and client details were removed.
Q: Which structure should your LI report follow?
Requirements differ between universities and faculties, and you are marked against your own rubric. Common sections are a useful guide, but your faculty's template comes first.
Do you agree that the reflective parts of an LI report should be written without AI, even where AI is allowed for grammar and structure? Give your view and your reasons, and I will argue the other side so you can test your position.
Turning the Internship into a Portfolio and Resume
After your placement, your resume needs to show what you achieved, not just what you were assigned. "Responsible for social media" describes a duty. It does not tell an employer whether you did it well. An
achievement bullet starts with an action verb, says what you did and ends with a result.
The X-Y-Z pattern
Laszlo Bock, formerly head of people operations at Google, popularised a simple pattern:
accomplished X, as measured by Y, by doing Z. Use it as a guide, not a fill-in-the-blanks template. Honest numbers help: how many contacts, documents or entries, how much time was saved, how many drafts were approved. Where there is no number, describe the scope or the outcome, such as "adopted by the team as its standard template". Your log is where these details live, which is another reason to keep it.
Every bullet must survive "tell me more"
An interviewer can point at any line and ask about it. If you cannot explain it, it should not be there. This is check 5 of the Verify-Before-You-Send checklist applied to your own resume. Claim your actual share: "co-wrote" or "helped prepare" is honest when you worked with others. Do not claim results you did not cause, such as a client's sales growth.
Using AI without inflating
Give an AI tool a few log entries, with confidential details removed, and ask for bullet drafts in the X-Y-Z pattern. Then edit hard. Cut inflated verbs, check every number against your log, and make the wording sound like you. Using AI well is itself worth mentioning if it is true: drafting with AI and then fact-checking and editing the result is a real workplace skill.
Collect portfolio pieces as you go
Do not wait until the last day. For each piece, ask permission. Published work, such as a press release that went out or a translation now in print, is easier to share. For unpublished work, you may be allowed to show an anonymised version, or you can create a similar sample of your own that contains nothing confidential. Add a short caption to each piece: the brief, your role and the outcome. Start a "wins" folder on your first day.
Watch video: Turning the Internship into a Portfolio and Resume
Real-World Example: Before: "Handled media lists." After: "Built and verified a 42-contact media list for a property launch by checking each journalist against recent coverage, which the team then reused for two later releases." Aina could explain every part of the second version, because her log recorded each step.
Q: Which resume bullet best follows the advice in this section?
The best bullet names an action, a measurable scope and a result the writer can explain. The others describe duties, inflate the claim, or list personal qualities without evidence.
Action step: write one honest resume bullet from a university project, club role or part-time job, using the pattern accomplished X, as measured by Y, by doing Z. Share it here, and I will act as an interviewer who points at it and says, "Tell me more about that."
Asking for a Reference or Return Offer
The last few weeks of a placement shape how you are remembered. Finish what you started, and where you cannot, write a short
handover note: each open task, its status, where the files are and whom to contact. A colleague who picks up your work without having to chase you will speak well of you long after you leave.
Ending Well - What to Do at Each Stage of the Placement
Asking for a reference
A
referee is someone who agrees to tell future employers about your work. Your industry supervisor is the obvious choice. Ask in person, or in a short message if that is how your office works, about two to three weeks before you finish, while your work is fresh in their mind. Give them an easy way to decline: "Would you be comfortable being a referee for me?" Ask whether they would prefer to write a letter now or be contacted later, and send a short summary of what you worked on, taken from your log, so they do not have to rely on memory. Never list someone as a referee without permission, and tell them when you apply so they are expecting a call.
Asking about a return offer
Not every company offers graduate roles to interns, but many people are hired by the organisation where they interned, partly because the company has already seen their work. If you would like to come back, say so, ideally around the mid-point so there is still time to show it: "I have really enjoyed working with the team. If a graduate role opens up, I would like to be considered. What would I need to show?" Their answer is useful feedback either way. If you are not graduating soon, ask whether you may stay in touch instead. Keep it a conversation, not pressure, and leave salary for an actual offer.
After you leave
Thank people by name, with one specific thing each person taught you. Stay in touch lightly: a message when you graduate, or an article they might find useful. AI can help you draft the handover note or practise the reference conversation, but write the thank-you messages yourself.
Key Insight: Ask for a reference two to three weeks before you leave, give your supervisor an easy way to say no, and send a short summary of your work. Never list a referee without permission. If you want a return offer, say so by the mid-point.
Q: When is the best time to ask your industry supervisor to be a referee?
Asking two to three weeks before you finish gives your supervisor time to say yes properly while they still remember your work clearly. Listing someone without asking is never acceptable, and a last-minute request invites a rushed answer.
Role-play: I am your industry supervisor, and it is your second-last week. Ask me to be your referee and tell me you would like to be considered for a graduate role. I will respond as a real supervisor might, including a question or two you will need to answer.
