Google DeepMind Adds Invisible Watermarks to AI-Designed Proteins for Biosafety

What happened

Google DeepMind has introduced 'SynthID Bio', a groundbreaking method to embed invisible watermarks within proteins generated by AI, without affecting their biological function. This innovation allows for the tracking and identification of AI-created biological molecules, significantly bolstering biosecurity efforts against misuse.

Google DeepMind's SynthID Bio invisibly watermarks AI-designed proteins without breaking them, allowing tracing of AI-made biological molecules for crucial biosecurity and safety.

Why it matters: AI's biological creations are now entering a new era of accountability and traceability, enhancing global safety.

• Google DeepMind can now invisibly 'sign' AI-designed proteins, including those targeting viruses like COVID-19.
• The watermarked proteins function identically to unmarked versions, proving the technique works without disruption.
• This makes it significantly harder to secretly create dangerous AI-generated biological molecules without detection and accountability.

Read our summary: Google DeepMind Adds Watermarks to AI-Designed Proteins Without Breaking Them

Open the interactive story

More Neural Snaps · Neural Toons · The Neural Feed