AI advances from content generation to biological design and transparency measures

Researchers from Stanford and the Arc Institute have utilized genomic models, specifically Evo 1 and Evo 2, to design complete genomes for bacteriophages—viruses that infect bacteria. Moving beyond mere digital simulation, scientists synthesized hundreds of these AI-designed sequences in a laboratory setting. Of these, sixteen resulted in viable phages capable of infecting E. coli.

While the term “generating life” is nuanced due to the biological debate regarding whether viruses are living beings and the necessity of human intervention through DNA synthesis, the experiment marks a shift from AI generating content (text, images, video) to AI proposing functional biological instructions.

In a separate development in AI safety, Anthropic announced that its Claude models will incorporate invisible watermarks into AI-generated text to improve transparency. This move aligns with the European Union's AI Act and will be implemented globally for all models released after August 2, 2026. The watermarks are designed to persist even if content is copied and pasted. Anthropic also plans to use C2PA standards to provide provenance information for AI-generated images, videos, and audio.