How to use AI safely and avoid AI scams
Check AI answers before you rely on them, keep private details private and recognise fake voices, images and messages. For anyone who uses AI, or could be targeted by people who do.
Who it’s for: Everyone: students, parents, professionals and older adults.

AI can save you time, but it can also be confidently wrong. In 2023, New York lawyers were fined US$5,000 after citing court cases that ChatGPT had invented. The cases sounded real, with names and quotes, but nobody had checked them. This course teaches the simple checks that would have caught them.
You'll learn the signs of a made-up answer, which is known as a hallucination. You'll practise opening a cited link and confirming it says what the AI claims. You'll write a list of things you'll never paste into a chatbot, such as passwords, card numbers and other people's private details. You'll also find the setting that decides whether your chats are used to train the AI.
The second half of the course is about people who use AI against you. You'll learn to judge whether a photo or video is real. You'll also set a family safe word, so a panicked call in a loved one's voice can be checked. You'll learn to judge scam messages by what they ask you to do, not by their spelling. You'll even make an assistant argue against your own idea before you act on it. The course is for everyone, including students, parents, professionals and older adults.
By the end you’ll be able to…
- Verify an AI claim by opening and reading its cited source
- Decide what is safe to paste into a chatbot and what never is
- Set a family safe word that defeats voice-clone scam calls
- Make an assistant argue against your idea before you act on it
Short on time? Start with the essentials
These 4 lessons cover the most useful ideas in this course. Each one takes about a minute.
How do I spot AI making things up?
Recognise the signs of a hallucinated answer (precise details, no source, made-up citations) and decide what to check.
Does the source really say that?
Open a cited link and confirm the page exists, is reputable and actually says what the AI claims.
What should I never tell a chatbot?
Apply a never-paste list (passwords, card and ID numbers, other people's private data, confidential work files).
How do I spot a scam message now?
Judge a message by what it asks you to do (pay, log in, act urgently), not by its grammar.
Every lesson in this course
This course has 12 video lessons of about a minute each, from the basics to expert habits. Each one ends with a question, three takeaways and its sources.
Foundations 3 lessons
How do I spot AI making things up?Essential
Recognise the signs of a hallucinated answer (precise details, no source, made-up citations) and decide what to check.
“Lawyers have been fined for using court cases a chatbot made up.”
Does the source really say that?Essential
Open a cited link and confirm the page exists, is reputable and actually says what the AI claims.
“A link isn't proof. Click it and see what it really says.”
What should I never tell a chatbot?Essential
Apply a never-paste list (passwords, card and ID numbers, other people's private data, confidential work files).
“Some things should never go into a chatbot. Let's go through six.”
Building skills 5 lessons
Is AI learning from my chats?
Find and set the model-training and history settings in your assistant, and know the defaults differ by app.
“Your chats might be helping train the next AI model. Here's how to check.”
How can I check if something's true?
Use lateral reading: leave the page, search who is behind a claim and compare independent sources.
“Professional fact-checkers often leave a web page almost straight away. Here's why.”
Is this photo or video real?
Check where a surprising photo or video came from, using a reverse image search, before believing or sharing it.
“An extra finger used to give away a fake photo.”
Could that call be a cloned voice?
Set a family safe word and a call-back rule so an urgent call in a loved one's voice can be verified.
“A few seconds of your voice can be enough to clone it.”
How do I spot a scam message now?Essential
Judge a message by what it asks you to do (pay, log in, act urgently), not by its grammar.
“Scam emails often used to be full of typos. AI is fixing that.”
Advanced 3 lessons
Does AI treat everyone the same?
Run the same prompt with only a name, gender or location changed and compare the answers for unfair differences.
“Change one name in your prompt. Does the advice change too?”
Who owns what I make with AI?
Explain that work made only by AI may not be protected, but your own creative changes can be.
“In the US, an image made only by AI can't be copyrighted.”
Is AI just telling me what I want?
Spot when an assistant just agrees with you, and get an honest view by asking it to argue the other side.
“Ask “are you sure?” and some chatbots will change a right answer.”
Expert 1 lessons
Can a web page trick my AI?
Explain how hidden instructions in a web page or email can hijack an AI tool, and keep it from acting without your OK.
“A hidden line on a web page can give your AI orders.”
Apps we recommend
These are the tools this course uses, and what each one is best for. We have no partnership with any of them.
Perplexity
Best for: Practising source checksEvery answer lists numbered sources, which makes it quick to practise checking claims.
FreemiumContent Credentials Verify
Best for: Checking where an image came fromUpload an image to read its Content Credentials, including whether an AI tool made or edited it.
FreeHave I Been Pwned
Best for: Checking if your email was in a data breachScammers use leaked data to make AI-written scams convincing. See what's already out there about you.
FreeGive your own AI use a safety check-up
Review how you and your household use AI. Check the privacy settings in each assistant you use, fact-check one answer you recently relied on, write your never-paste list, agree a rule for urgent money or login requests, and make an AI argue against one of your decisions.
What you’ll make: A one-page personal AI safety plan: settings chosen, never-paste list, scam rule and one worked fact-check.
Check your own work
- Training and history settings are checked and recorded for every assistant you use
- One AI answer is verified against two independent sources, with links
- A never-paste list of at least 5 items that fits your own life or job
- A call-back or safe-word rule for urgent requests is agreed with at least one other person
- One decision was tested by asking the AI to argue the other side, and what changed is noted
Further reading and viewing
We’ve chosen these articles, research papers, videos and guides from other trusted sources.
Scammers use AI to enhance their family emergency schemes
How voice-clone 'family emergency' calls work and how to verify before you pay.
AI and cyber security: what you need to know
Plain guidance on AI risks, including what not to share with AI tools.
AI Risk Assessments
Independent safety ratings of popular AI products for kids and teens.
Copyright and Artificial Intelligence
Official reports on AI training, copyrightability of AI output and digital replicas.
Managing parental controls in ChatGPT
Step-by-step guide to linking a teen account and choosing its settings.
Questions about this course
How can I spot an AI scam?
Look at what the message or call asks you to do, not how polished it sounds. AI now writes scams with perfect spelling, and it can copy a familiar voice. Be wary of anything urgent that asks you to pay, log in or share a code. Hang up or close the message, then contact the person or company on a number you already know.
What is a voice-cloning scam?
It's a call that sounds like someone you love, often in a panic and asking for money. AI can copy a voice from a short recording, such as a video posted online. The best defence is a family safe word that only you and your relatives know. If the caller can't give it, hang up and call them back on their usual number.
What should I never tell a chatbot?
Keep passwords, card and bank details, and ID numbers out of every chatbot. Don't paste other people's private information, such as a friend's medical news. Leave out confidential work files too, unless your employer has approved the tool. Some assistants use chats to improve their models, depending on your settings, so assume anything you type could be stored.
How can I tell if a photo is AI-generated?
Visual clues like odd hands or garbled text help less than they used to. Instead, check the context. Ask who posted it first, and run a reverse image search to find the original. Some images also carry Content Credentials, which can show whether an AI tool made or edited them. If you can't confirm where it came from, don't share it.
How long are the lessons?
Each lesson takes about a minute. You watch a short captioned video, answer one question and get three takeaways with the sources. Some lessons also let you practise on a realistic situation and get feedback. Finished lessons come back as quick one-question reviews a day later and then a week later, so the habits stick.
Get early access to this course
Tau opens soon. Join the waitlist and we’ll email you as soon as you can start this course.
We’ll only email you about Tau, and you can unsubscribe any time. Privacy policy.
We’ll email you as soon as Tau opens.


