AI Designs First Living Viruses from Scratch, Some Outperform Natural Ones
新智元 · wechat · 2026-08-08
A joint research effort by Stanford University and the Arc Institute, published in Science, has successfully used the generative AI model Evo to design fully functional, living viral genomes from scratch for the first time.
The model generated 700,000 candidate genomes, from which 285 were selected for DNA synthesis and introduced into E. coli. Ultimately, 16 new viruses survived and replicated, with some AI-designed variants lysing bacteria even faster than their natural counterparts.
Experiments also showed that AI-generated bacteriophages could rapidly break through bacterial strains resistant to natural phages. This marks the dawn of generative genome design, drastically compressing the timeline for developing new antimicrobial agents from years to a single inference run.
More from Research
- PrimeIntellect verifiers v0.3.1: Model Interception and Persistent ACP Sessions — xeophon · 2026-08-27
- RAG Isn't Dead: Navigating Retrieval vs. Agentic Search — hugobowne · 2026-08-27
- Podcast: What happens when you let an AI run a science lab — JMarty97 · 2026-08-27
- AI audit of 19 papers: 97.7% of comments identified real issues — littmath · 2026-08-27
- Mathematician Litt: Paper Errors Mostly From Badly Propagated Edits, Not Deep Flaws — littmath · 2026-08-27
- Terence Tao ran all his published papers through AI error-finding scaffolds — littmath · 2026-08-27