Has Google’s AI watermarking system been reverse-engineered?
2026-04-14 · The Verge
Google's supposedly tamper-proof AI image watermarking system (SynthID) may have been reverse-engineered by a developer who claims to have figured out how to strip the watermarks or forge them. Google disputes the claim, but the open-sourced proof-of-concept just landed on GitHub—so much for AI safety theater.