Latenode

Anthropic says Claude made 17 scorpion-toxin binders, not antivenom

The result is a useful checkpoint for AI-assisted biology: binding data can be real progress without establishing neutralization, animal efficacy, or a medical use claim.

Anthropic says Claude made 17 scorpion-toxin binders, not antivenom

Anthropic says Claude helped create 17 protein designs that met its criteria for binding dangerous scorpion toxins, but the company does not claim they work as antivenom.

The Anthropic research says Claude generated 318,484 candidates. Adaptyv Bio and Twist Bioscience tested 211, with qualifying binders found for four toxins. None qualified for Ts1, and designs aimed at multiple toxins failed to bind more than one.

The main limit is neutralization. Anthropic tested binding to recombinant toxins, not whether the proteins stop venom in cells or animals. Its strongest candidate showed single-digit nanomolar affinity or better and competed with a known neutralizing antibody, but that is still an early lab signal rather than evidence of an antidote.

When models and APIs shift keep the workflow running

Connect the models and apps from stories like this in one scenario. If a vendor changes routing, pricing, or availability, you update the workflow — you don't start over.

Start Free

Free forever plan. No credit card required.