Updated
Updated · The Verge · Jul 17
TikTok Tests AI Deepfake Detection Tool for Some US Creators as YouTube Broadens Similar Access
Updated
Updated · The Verge · Jul 17

TikTok Tests AI Deepfake Detection Tool for Some US Creators as YouTube Broadens Similar Access

3 articles · Updated · The Verge · Jul 17

Summary

  • TikTok has begun testing an opt-in tool with some US creators that scans for AI-generated likenesses and lets them report suspected unauthorized deepfakes.
  • Creators in the test must verify their identity through Jumio using a real-time selfie and ID check before TikTok’s system can match their likeness against AI-generated content.
  • TikTok says it does not retain ID documents and uses facial information only for likeness matching and to help identify potential misuse.
  • The test puts TikTok alongside YouTube, which has been developing a similar detection system and recently opened it to all adult users.

Insights

With its tech partner losing a major patent suit, can TikTok's new anti-deepfake tool actually work?
Does an 'opt-in' system absolve platforms of their legal duty to police all AI-generated fakes?
Is trading biometric data for protection a fair price for creators to pay for online safety?