AI risks: privacy, bias, deepfakes and scams

Know where your data goes, spot biased or false AI output, stop deepfake scams, and write the AI rules your household and team follow.

You probably paste things into an AI assistant every week without knowing where they end up. Your parents may get a call one day in a voice that sounds exactly like yours, asking for money. Your manager may ask whether the team is allowed to use AI on client work, and nobody has an answer. This course covers the risks that come with AI for ordinary users and employers, from the PDPA and bias to copyright, cloned voices and false information, and ends with rules you write down and use.

What you'll be able to do

• Check an assistant's settings and account type and say where your conversations are stored and how they may be used
• Decide whether a work use of AI involves personal data and what the PDPA asks of you
• Test an AI output for bias by changing names, genders or other details and comparing the results
• Explain why ownership of AI-generated content is unsettled and apply sensible rules at work
• Set up a family safe word and callback routine that defeats a voice-clone or deepfake call
• Check a claim, image or video against its original source before sharing it
• Write a personal AI policy and a one-page team AI policy

Syllabus

Module 1: Follow your data after you press send

What happens to your prompts, files and chat history, how account type and settings change that, and how to sort information into what is safe to share and what never goes in.

• 1.1 What happens to the words you type into an assistant
• 1.2 Personal, business and enterprise accounts are not the same
• 1.3 History, memory, temporary chats and deletion
• 1.4 Audit the AI tools you used this month

Module 2: Use AI at work without breaching the PDPA

The PDPA obligations that apply when personal data goes into an AI tool, what the PDPC has said about AI systems in general terms, and how to check a work use before you start.

• 2.1 PDPA basics every employee should know
• 2.2 When pasting a customer list becomes a disclosure
• 2.3 What the PDPC has said about AI systems
• 2.4 Run a PDPA check on three work uses of AI

Module 3: Spot bias before it reaches a decision

Where bias in AI systems comes from, what well-documented cases teach, and a simple swap test you can run on any assistant output that affects people.

• 3.1 Where bias in AI comes from
• 3.2 Documented cases: hiring, faces and risk scores
• 3.3 The swap test for everyday AI output
• 3.4 Bias-test an AI task you rely on

Module 4: Know who owns what AI makes

Why copyright in AI-generated content is unsettled and varies by country, what risks come from inputs and outputs, and the working rules a careful team follows meanwhile.

• 4.1 Who owns AI output is still being decided
• 4.2 Risks on the way in and on the way out
• 4.3 Working rules for AI content at work
• 4.4 Write a content use note for your team

Module 5: Stop deepfake video and cloned-voice scams

How voice cloning and face-swapped video are used to impersonate family, bosses and public figures, the documented cases, and the safe word and callback routine that defeats them.

• 5.1 How a cloned voice or a face swap is made
• 5.2 Real cases: the Arup video call and fake leaders
• 5.3 Safe words, callbacks and questions only they can answer
• 5.4 Set up your family safe word and callback plan

Module 6: Check what you read before you share it

How AI makes false text, images and sites cheap to produce, why assistants invent sources, and the lateral reading and image checks that confirm a claim before you pass it on.

• 6.1 How AI makes false information cheap
• 6.2 Lateral reading: leave the page to check it
• 6.3 Checking images and video
• 6.4 Fact-check five claims before you forward them

Module 7: Keep your own skills sharp while you use AI

Why people accept confident machine output too easily, which skills fade when you stop practising them, and how to decide what to hand to AI and what to keep doing yourself.

• 7.1 Why we trust confident machines too much
• 7.2 Skills you lose when you stop practising them
• 7.3 Decide what to hand over, what to assist, what to keep
• 7.4 Map your tasks: delegate, assist or keep

Module 8: Write the AI rules for yourself and your team

Turn the course into two short documents: a personal AI policy you follow and a one-page team policy covering approved tools, data, review, disclosure and what to do when something goes wrong.

• 8.1 What goes in a personal AI policy
• 8.2 What goes in a team AI policy
• 8.3 Getting a policy used, not filed
• 8.4 Draft your personal and team AI policies

Frequently asked questions

How long does the course take?

About eight hours across eight modules, including the exercises. The family safe word exercise needs a real conversation, so plan time for it.

Is this legal advice on the PDPA or copyright?

No. The course gives working knowledge for everyday use. For a specific case, speak to your data protection officer, your legal team or a lawyer, and read the current PDPC guidance yourself.

How is this different from Scam-proof your money?

Scam-proof your money covers the common scams and the bank and phone defences against them. This course covers the risks that come from AI itself, including deepfake and cloned-voice scams in depth.

Do I need technical knowledge?

No. You need to have used an AI assistant a few times. AI fundamentals: what it is, how it works, where it fails explains the technology if you want the background.

Explore more

• All Academy courses