Verify and don't get fooled
The warning signs to spot, how to cross-check sources, and what to do about manipulated content.
What you'll be able to do
- Spot the warning signs of dubious content
- Cross-check information with another source
- React correctly to a suspected deepfake
Program — 8 steps
- 1In 2023, a lawyer submitted court decisions to a tribunal that were entirely invented by ChatGPT. The result: sanctions, and a textbook case.
- 2An AI answers with total confidence — true or false, the tone never changes.
- 3True or false?
- 4Warning sign or reliable source?
- 5What is the C2PA standard, launched in 2021, for?
- 6Tap the passage that describes a reliable warning sign
- 7Ask the same important question to two different AIs, and see which one cites a verifiable source.
- 8Hallucination Hunt
Recommended prerequisite : Agents and automation
Your progress stays on this device — no account, no email.
On this page
1. Verify and don't get fooled
- A magnifying glass over a green checkmark, symbolizing verification.
In 2023, a lawyer submitted court decisions to a tribunal that were entirely invented by ChatGPT. The result: sanctions, and a textbook case.
- A scale weighing a warning sign against a reliable source.
An AI answers with total confidence — true or false, the tone never changes.
Spot the signs: a reference found nowhere else, a very precise number with no source, an overconfident tone on a niche topic.
Before any important decision (health, legal, money), always cross-check with a reliable, independent source.
A generative AI can state false information with the same confidence as true information.
A very confident tone on a niche topic guarantees the answer is reliable.
Deepfakes today are still always easy to spot by eye.
The C2PA standard is used to track a digital content's origin and edits.
Getting fooled once by AI-invented information is rare and only happens to careless people.
True or false?
Warning sign or reliable source?
A reference found nowhere else
A very precise number with no source cited
A very confident tone on a niche or very recent topic
The World Health Organization's official site
A reference document cited with a verifiable link
What is the C2PA standard, launched in 2021, for?
Tap the passage that describes a reliable warning sign
An answer written in a calm, measured tone is always guaranteed to be accurate. A citation or reference that can't be found anywhere else should raise suspicion. Very precise numbers given without any cited source deserve to be checked.
Ask the same important question to two different AIs, and see which one cites a verifiable source.
- A magnifying glass over a green checkmark.
Hallucination Hunt
"According to a 2019 study by Dr. Martin Dubois in the Journal of Advanced AI Research..." — can't find it anywhere else.
"According to the World Health Organization's official site, here is the recommendation..."
An answer gives a very precise number ("exactly 73.4%") with no source cited at all.
For an important health decision, trusting a single AI answer without checking elsewhere...
Hallucination Hunt
12 seconds per round: spot whether the statement contains a warning sign or seems reliable.