Skip to content

News

Claude designs protein binders against scorpion neurotoxins

Working inside Claude Science, an internal Anthropic research model designed protein binders against five scorpion neurotoxins. Of 211 designs tested in two outside labs, 17 bound their target; Anthropic is releasing all designs and data but has not tested whether they neutralize venom.

On 20 October 2026 Anthropic's science team described a project in which an internal research model, running inside Claude Science, designed small protein binders, or minibinders, against scorpion neurotoxins. The prompt was adapted together with the investment research firm Citrini Research, and the model worked largely on its own for 64 hours, steered by four follow-up prompts. It targeted five neurotoxins from scorpions in Mexico, Brazil, North Africa and the Middle East, aiming at the part of each toxin that attaches to sodium channels in nerve cells.

Claude generated 318,484 designs with 12 design tools and picked 180 for testing; the team added 31 more. Adaptyv Bio and Twist Bioscience tested all 211 in their labs. Seventeen met Anthropic's binder criteria, each against a single toxin, and none of the designs aimed at several toxins at once bound more than one. The two strongest bind the Mexican scorpion toxin Cn2 with single-digit nanomolar affinity and likely share a binding site with an antibody reported to neutralize it in mice. The campaign used on the order of 10,000 GPU hours.

Anthropic stresses that the binders were not tested for neutralization in cells or animals, so it does not claim to have produced an antivenom. All 211 designs and their data are public, alongside a technical report.

Sources

The headline and summary are ours, drafted with AI from the sources above and checked automatically before publication. See how we use AI.