Verified by external labs, the AI’s designs matched or exceeded the performance of the best previously published results, often completing tasks in 48 hours that typically require weeks or months of specialized human effort.
These results matter because accelerating protein design could drastically shorten the early stages of drug discovery, where identifying molecules that bind to disease-related proteins is a primary bottleneck.
By automating the orchestration of complex "wet lab" (physical biological testing) data and specialized computational folding models, Claude demonstrated it can handle the sophisticated scientific reasoning usually reserved for PhD-level experts.
Furthermore, in a separate chemistry test, a generally available version of Claude analyzed raw laboratory data to identify compound purity and molecular structure in under 25 minutes, a task that typically involves manual processing by chemists using proprietary software.
To support further research, Anthropic plans to launch a dedicated access program for scientists to use these advanced capabilities.
The company noted that while these tools can speed up the development of life-saving therapies, they are considered "dual-use" technology, meaning they could potentially be misused for harmful biological research.
Consequently, these specific scientific features remain restricted in general-access versions of the model as Anthropic develops safety protocols and trusted access frameworks for the research community.