Anthropic launches beta Life Sciences Verification Program for verified biology teams

The program’s safeguards target three threat models—access compromise, insider threats, and agent misuse—by tying model access to declared use cases and monitoring traffic offline for patterns of abuse. Data from flagged activity is retained for 30 days, isolated from training pipelines, and can be reviewed by organization admins. Anthropic cited a recent threat‑intelligence report that documented several dual‑use misuse attempts, arguing that verified, monitored access is essential for safely offering frontier AI capabilities to drug‑discovery, clinical development and manufacturing teams. Early adopters such as Xaira, Edison and Manifold Bio have already pledged to integrate the models into their pipelines.