AI, deepfakes & synthetic harm
AI-generated profiles, voice clones, deepfake nudes and automated grooming bots running at industrial scale. The fastest-growing category of harm and the one least covered by existing platform policies.
Scope of this assessment area
Generative AI has collapsed the cost of producing convincing intimate content of real people. Platforms are now the first line of defence against non-consensual synthetic imagery, and the policies have not caught up with the threat.
What we look for
Observable patterns that trigger investigation under this area.
Criteria a platform is scored against
Each criterion is publicly defined and applied consistently across every platform we review.
Synthetic-media detection at upload
Watermark / fingerprint checks on every uploaded image.
Consent verification for likenesses
Proof that depicted identity has consented.
Takedown speed for NCII
Hours-to-removal, hash-based blocking across the network.
Hash-sharing participation
Active membership of StopNCII / NCMEC / GIFCT-equivalent hash networks.
Clear policy on AI-only personas
Disclosure required; revenue and prominence limited where warranted.
If you are affected - what to do first
Sequential, low-risk actions you can take right now.
Report with the deepfake flag
Use the synthetic-media path, not the generic copyright path.
Submit hashes to StopNCII.org
Hashes propagate to participating platforms automatically.
Contact the platform's NCII team
Most large platforms now have a named NCII contact.
Consider legal action
Many jurisdictions now have deepfake-specific criminal and civil routes.
Have evidence in this area?
Submit confidentially - encrypted at rest, named reviewers only, you can submit anonymously.