TL;DR
Linus and the team decided to revisit their deepfake experiment from five years ago—only this time it was shockingly easier. They show how they trained an AI model to mimic Linus’s face and voice (with plenty of hiccups syncing audio and lining up footage), then test whether average viewers—even older relatives—can spot the fake.
Along the way they share practical tips for detecting AI-generated video (and point to Hany Farid’s work on spotting fake photos), bump into the usual challenges of DIY deepfakes, and remind you that these tools are improving faster than ever. Sponsored by Bitdefender, the video comes loaded with chapter markers and extra resources for anyone who wants to learn more.
Watch on YouTube
Top comments (0)