Lesson 4 of 7
Deepfakes & misinformation
6 min read
A clip of a world leader saying something shocking spreads to millions before lunch. It never happened. When anyone can fake anything, how do you know what's real?
Synthetic media at scale
AI can now generate a believable photo, clone a voice from seconds of audio, or make a deepfake video — cheaply, instantly, and in bulk. That's a gift for scams and propaganda: a fake that once took a studio now takes a prompt. The danger isn't one perfect fake; it's the flood of good-enough ones, and the doubt it casts on real footage too.
Fakes are now cheap, fast, and convincing — the volume is the real threat.
Why glitch-spotting fails
The old advice — look for weird hands, warped text, robotic voices — is expiring fast as models improve. Trusting your eyes is exactly the instinct that gets people fooled. The durable move is provenance: check where a file came from, look for Content Credentials or watermarks, and cross-check with a trusted source before you believe or share it.
Don't judge a fake by its pixels; judge it by its source and its record.
Before you share a shocking image or clip, pause: where did it first appear, does a reliable outlet report it, and does it carry any provenance marks? Slow is the new sceptical.
The takeaway
- —AI makes convincing fake media cheap, fast, and plentiful.
- —Spotting glitches no longer works — models have gotten too good.
- —Check provenance and cross-check the source before believing or sharing.
What's the reliable way to handle a shocking viral clip?
Continue in the app
Take the whole AI, Jobs & Society course — tracked
Get your personalized path, progress and streaks in the app — this lesson and every next one, in order.