Copy article

YouTube's new likeness detection tool hiding small print privacy concerns

ended 03. December 2025

YouTube's new likeness detection tool claims to protect creators from AI deepfakes, but the fine print reveals something more concerning: by uploading government ID and facial biometrics to fight unauthorised AI use of their likeness, creators are simultaneously granting Google permission to train AI models on that same biometric data. 

The policy explicitly states that public content, including biometric information, can be used to help train Google's AI models and build products and features. 

It's a perfect illustration of how the AI industry keeps asking humans to sacrifice privacy and agency for protection they shouldn't need in the first place. 

The tool launched this week to millions of creators in the YouTube Partner Program, requiring them to submit government ID and a selfie video performing randomised actions to verify their identity. 

Meanwhile, Google's own Veo 3 video generation tool was trained on a subset of the more than 20 billion videos uploaded to YouTube, likely including hundreds of hours from the very creators now being asked to hand over their biometric data to fight the deepfakes this training enabled. 

We want your views:

  • Should biometric data collected for identity verification be explicitly prohibited from AI training under any circumstances? 
  • What does it mean for human agency when the only way to protect your likeness is to hand over even more sensitive data to the same ecosystem that created the problem? 
  • Are we normalising a future where every interaction with AI systems requires sacrificing privacy just to maintain basic digital dignity? 
  • How should we measure the true cost of "free" AI safety tools when the price is paid in biometric data and future model training? 
  • What would genuinely transparent, human-centred deepfake protection look like if designed without conflicting commercial incentives?

 Responses by 8am tomorrow.

1 responses from the Newspage community

Copy all

Copy

Colette Mason
AI Ethics Consultant at Clever Clogs AI
What creators actually need: biometric systems with cryptographic guarantees that data cannot be repurposed. Not "we promise not to," but "we technically cannot." Until that exists, every ID upload is a bet that corporate ethics survive quarterly earnings pressure. The real cost isn't data harvested, it's the agency surrendered, one "necessary" compromise at a time.