YouTube Opens Deepfake Detection to Everyone as Lawsuits Land

YouTube is expanding its AI likeness detection tool to all users over 18, The Verge reports. Anyone can now upload a selfie, have the platform scan for videos that appear to use their face, and flag matches for review. The feature was previously limited to creators, politicians, journalists, and other public figures.

On the same news cycle, The Verge also reports that Snap, YouTube, and TikTok settled a landmark lawsuit brought by Kentucky's Breathitt County School District alleging social media addiction harmed student learning and mental health. Terms were not disclosed. Meta did not settle and still faces trial. More than 1,000 similar suits are queued behind this one.

Reading these together clarifies what is actually happening. Platforms are not having a values awakening. They are pricing in liability.

Two tracks, one driver

                  Platform behavior change
                          ▲
                          │
         ┌────────────────┴────────────────┐
         │                                 │
  Litigation pressure              Regulatory pressure
         │                                 │
   ┌─────┴──────┐                  ┌───────┴────────┐
   ↓            ↓                  ↓                ↓
School      AI likeness        EU DSA / UK         State age
district    & deepfake         Online Safety       verification
suits       suits              Act                 laws
   │            │                  │                │
   ↓            ↓                  ↓                ↓
Snap/YT/TT  YouTube opens     Mozilla pushes      Platform ID
settle      detection         back on VPN         checks expand
            to all adults     restrictions

The likeness tool is genuinely useful. Ordinary people get impersonated in scam ads, romance fraud, and political smears, not just celebrities. Giving everyone a self-service way to scan for deepfakes of themselves is overdue. But the structural reason it ships now, rather than two years ago when the technology was already there, is that the legal exposure for hosting non-consensual AI likenesses has grown sharp enough to overtake the engineering cost.

The school settlement carries the same logic. A district in rural Kentucky extracted concessions from three of the largest platforms in the world. The settlement amount is private, but the precedent is not: a small plaintiff with a coherent harm theory can force a deal. Plaintiffs' lawyers running the other 1,000 cases now know what the floor looks like.

The wider regulatory context matters too. Mozilla this week urged UK regulators not to weaken VPN services as part of the country's online safety regime, arguing they are essential privacy tools. The UK, the EU, and a growing list of US states are all pushing platforms toward more identity verification, more content scanning, and more liability for what users post and what AI generates. Each new obligation gets met with a feature rollout that looks like user empowerment and functions like a compliance artifact.

The deepfake selfie scan will help real people. It will also generate a database of verified faces tied to YouTube accounts, sitting inside Google. Whether that tradeoff is worth it depends less on Google's intentions than on what the next subpoena, breach, or policy change does with the data. The companies building these tools under legal duress are not the ones who get to answer that question.