Unit 3 · Level 5 · AI × crypto
AI-powered scams
Scammers now run deepfake video livestreams of famous CEOs promoting 'giveaways', clone voices from seconds of audio to fake a family emergency, and send phishing messages polished by AI in perfect Dutch, German, or Polish. In early 2024, a finance worker in Hong Kong wired about $25M after a video call where every 'colleague' was a deepfake. Old rule, new stakes: verify the request, not the face.
Free to play. No ads, no token, no account needed to start.
What you get asked
Why did 'it looked and sounded exactly like him' stop being evidence of anything?
A few seconds of audio and some photos are enough. Identity must now be verified through a separate channel you initiate, not through what your eyes and ears receive.
A livestream shows a real tech billionaire announcing: 'Send 1 ETH, receive 2 back to celebrate our launch!' What is this?
'Send X, get 2X back' has been the scam template for a decade; deepfakes just gave it a famous face. Nobody doubles strangers' money; the sent ETH is gone the moment it confirms.
Match the AI-era scam to the defence that beats it
Notice every defence is procedural: callbacks, bookmarks, hard rules. Procedures don't get fooled by production quality; eyes and ears do.
No matter how convincing the caller, video, or 'support agent': never share or type your ___ anywhere someone directs you to.
League 1's iron rule survives every technology shift: the seed phrase never leaves you. Your public address is fine to share; the twelve words are not. No legitimate party ever needs them.
AI made scams cheaper, fluent, and scalable. What's the correct strategic response?
Broken grammar and glitchy videos were yesterday's tells, and they're gone. What survives: you initiating contact through known channels, hardware wallets, and rules you never break under pressure. 🐜
The rest of this unit
Agents with wallets, decentralized compute, narrative coins, and the deepfakes hunting your keys.