DeepMind's SynthID Bio watermarks AI-generated proteins without breaking function
mikeflache · x · 2026-10-08
Google DeepMind researchers (Demis Hassabis, David Stutz et al.) published SynthID Bio in Nature: a framework embedding invisible cryptographic signatures directly into AI-generated protein sequences and predicted 3D structures, without harming function.
It extends provenance tracking from digital pixels to the building blocks of life, serving as a biosecurity layer that prevents AI-generated toxins from bypassing DNA synthesis screening while keeping public databases free of unverified computational slop. Open access; 41k views already.
More from Safety
- Corridor hosts NYC meetup next week on securing agentic coding — jefrankle · 2026-10-08
- NSA spending billions of dollars a year testing frontier AI models, sources say — coherence · 2026-10-08
- Account hijacked despite 2FA: victim urges passkeys, as teortaxesTex pitches always-on agents for email alerting — teortaxesTex · 2026-10-08
- Humans can't scale AI policy alone; agents must help build institutions — jacyanthis · 2026-10-08
- Anthis: human–AI relationship case studies are today's most important data — jacyanthis · 2026-10-08
- Epistemic re-entry: local competence doesn't guarantee end-to-end corrective competence in AI — blessedfortherest · 2026-10-08