● Covered by 1 source · 1 reportMedium impact1 neutral
Google DeepMind Develops Watermarking for AI-Designed Proteins
Key points
- Google DeepMind developed a protein watermarking system.
- Watermarks are embedded in protein sequences without functional impact.
- The system helps identify AI-designed proteins from trusted sources.
- Based on Google's SynthID technology for AI-generated content.
Addressing Biosecurity Concerns in AI-Designed Proteins
AI tools are increasingly used to design proteins, leading to advancements like plastic-digesting enzymes and venom blockers. However, these same tools could be misused to create toxins or modify viral proteins, posing biosecurity risks. Existing software for identifying threatening DNA sequences does not recognize AI-designed proteins, creating a gap in security.
Introducing Protein Watermarking
Google's DeepMind team has proposed a solution: protein watermarking. This system embeds a watermark directly into the protein sequences during their AI-driven design. The watermark does not compromise the protein's intended function.
How the Watermark Functions
The watermarking allows researchers to identify proteins designed by trusted sources. This distinction enables closer scrutiny of all other AI-designed proteins, enhancing biosecurity measures. The technology aims to differentiate between beneficial and potentially harmful AI-generated biological material.
Underlying Technology: SynthID
This protein watermarking system is based on Google's SynthID technology, which is used for watermarking AI-generated digital content like text and images. SynthID subtly influences the AI's choices, distributing a systematic bias throughout the output. This makes the watermark difficult to remove without knowledge of its encoding, and it can survive basic modifications like exporting or resizing.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
The daily brief
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Today's brief
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's brief
New every morning, and the back catalogue is archived by date.
Reporting from
Google's DeepMind team published research on a protein watermarking system to identify AI-designed proteins. This system embeds a watermark directly into protein sequences without affecting their function, allowing for the distinction between trusted and potentially threatening AI-generated proteins.