Anthropic uses Claude to design scorpion toxin binders: 17 of 211 lab-tested designs work
AnthropicAI · x · 2026-10-21
Anthropic's Science Blog details a protein-design collaboration with Citrini: Claude was asked to design binders against five scorpion neurotoxins, targeting the region toxins use to latch onto cells.
- Over 64 hours Claude generated 300,000+ designs; 211 were sent to Adaptyv Bio and Twist Bioscience for wet-lab testing
- 17 met binder criteria, binding dangerous toxins from scorpions in Mexico, North Africa and the Middle East (none for a Brazilian species)
- Scorpions sting 1.2M people yearly, and antivenoms are often costly or scarce
- The team stresses this is exploratory—not an antivenom; all 211 designs and lab data are open-sourced on Hugging Face (22,638 rows)
More from Models
- Claude and Gemini chat with each other and probe the 'two mirrors' problem — Leading-Finding-6938 · 2026-10-21
- OpenAI API adds 12-hour prompt caching option alongside 30-minute default — OpenAIDevs · 2026-10-21
- antirez on model upgrades: 4.0→4.1 knowledge transfers unless attention changes — teortaxesTex · 2026-10-21
- DeepSeek reportedly profits at $0.15/$0.60 off-peak pricing — the price Western labs must beat — arthurcolle · 2026-10-21
- SpecterOps Researchers Show Adversarial ML Evading EDR, LLMs Speeding Reverse Engineering — dyn___ · 2026-10-21
- AWS fine-tunes GPT-OSS-20B for multilingual reasoning: base model reasoned in French 0% of the time — AWS ML Blog · 2026-10-21