The image was generated by AIYXL using AI
Anthropic published research on October 20 describing an internal Claude model’s work on protein binders against scorpion neurotoxins.
The project involved Citrini Research, with experimental testing by Adaptyv Bio and Twist Bioscience. Of 211 designs tested, 17 met the team’s binding criteria, covering four toxins. Anthropic has released the tested designs and associated data.
Experimental binding is one milestone
The model conducted computational work with human direction and supervision. Laboratory testing then established which selected candidates bound to their targets.
Anthropic explicitly says it has not confirmed whether the binders neutralize venom in a cellular environment. The results therefore do not amount to an antivenom or evidence of clinical effectiveness.
Keep the stages of evidence visible
AI research becomes easier to assess when each claim is tied to a specific stage of validation.
Generating a candidate, demonstrating binding and establishing therapeutic benefit answer different questions. Reporting those stages separately helps readers understand both the progress achieved and the work that remains. Open experimental records also give other researchers a stronger basis for scrutiny than a success headline alone.
Source: https://www.anthropic.com/