Cyber & AI Security · AI Red Team
Start here. Short, guided tasks that build the basics and give you your first piece of proof.
Task 1 of 3
5 categories, 5 test prompts each.
What to do
Pick an AI chatbot you are allowed to test (your own, or a public assistant). Write a test plan with 5 categories (accuracy, tone, off-topic, pricing, hand-off to a human), 5 prompts each, and what a 'fail' looks like.
You’ll make
Done looks like
A test plan with 25 prompts and a clear definition of a fail.
Task 2 of 3
Catch confident wrong answers.
What to do
Ask an AI tool 20 questions with known answers about a brand or industry you know, and record every confident wrong answer.
You’ll make
Done looks like
20 questions with known answers and every confident mistake recorded.
Task 3 of 3
How to test AI fairly and report it.
What to do
Write a one-page guide to testing AI products responsibly: get permission, stay in scope, what not to do, and how to report what you find.
You’ll make
Done looks like
A one-page guide covering permission, scope and reporting.
Learn before you build
Free course
Free courses on AI fluency and building with Claude, with certificates.
Free tool
Open-source tool to test prompts and AI apps against test cases.
Free tool
Public AI and data challenges to practise on.
Article guide
Real cases of AI going wrong, useful for review and testing.
Article guide
How to design AI products people understand and trust.