AI Awareness Training

A free AI literacy course for verifying synthetic media

This self-paced AI literacy training teaches a practical verification workflow for AI-generated images: pause, question the context, run a reverse image search, treat detector and metadata results as signals, and check the source before you share or act. A closing deepfake-awareness chapter applies the same habits to AI video and voice, including verifying payment or access requests through a separate trusted channel. The course is free and requires no account.

Why "look closer" is no longer the lesson

A convincing generated image may not be identifiable by eye alone. In a 2025 study of 2,000 UK and US consumers, confidence in detecting deepfakes remained above 60%, while only 0.1% correctly distinguished all real and deepfake stimuli across images and videos. This strict all-items-correct result is not an average image-accuracy score. The course therefore starts by measuring how well looking actually works for you, then hands over a four-step behaviour guide and the tools that back it up — each chapter ends with a short picture check of what you would actually do, and the course ends with a recap.

Ten practical course chapters

• How hard is it, really? — five images, no context, no hints
• Don't share unchecked — rule 1: why a reaction, even a skeptical one, still spreads it
• The human eye — rule 2, tool 1: three pairs, find the real one, and which flaws you can actually verify
• Reverse image search — tool 2: the strongest single check, and how to run it
• AI detectors & metadata — tool 3: a signal, never a verdict
• Check the source — tool 4: account age, sender address, forwarding history
• Report misinformation — rule 3: using the platform's mechanism, not the comment section
• Respond factually — rule 4: how to correct something without it backfiring
• Excursion: AI video & voice — what changes beyond still images, including off-channel verification
• Recap: what you learned — every chapter in one line, next to your own results, plus the one sequence to keep

Learning objectives

1. 1. Treat looking as a weak first signal: name a flaw only when you can verify it, treat a hint as a reason to check, and draw no conclusion when there is nothing. Practised in chapter 1, 3 · scored by 2 picture questions in the chapter 3 check
2. 2. Pause before sharing or reacting to a claim nobody has confirmed. Practised in chapter 2 · scored by one picture question in the chapter 2 check
3. 3. Use reverse image search to find earlier, dated copies — and read the result correctly, including a real photo with a false caption. Practised in chapter 4 · scored by 2 picture questions in the chapter 4 check
4. 4. Read detector scores and Content Credentials as signals, never as verdicts. Practised in chapter 5 · scored by 2 picture questions in the chapter 5 check
5. 5. Check the source — account, sender domain, independent confirmation — before trusting a claim. Practised in chapter 6 · scored by one picture question in the chapter 6 check
6. 6. Report deliberate deception through the platform, in the category that fits. Practised in chapter 7 · scored by one picture question in the chapter 7 check
7. 7. Correct misinformation with a specific, checkable reason instead of a verdict — and label your own AI images. Practised in chapter 8 · scored by 2 picture questions in the chapter 8 check
8. 8. Verify urgent money, credential or access requests off-channel, whoever they seem to come from. Practised in chapter 9 · scored by one picture question in the chapter 9 check

For groups and classrooms

The course is built for one person at a time. To teach it live to a class — you present on the board, students answer the questions on their phones — use Classroom: AI Awareness Training. For just the image quiz as a class game, switch to Simple Quiz there. The measured background to the numbers used here is in our first-party analysis and the detector benchmark.

Questions about the course