Anthropic launches Life Sciences Verification Program

Anthropic has introduced the Life Sciences Verification Program (LSVP) in beta, designed to bypass the strict biological safeguards of its standard Fable models. The program grants verified teams and institutions access to the Mythos, Opus, and Sonnet models—specifically Mythos 5.1, Opus 5, and Sonnet 5—with refined classifiers tailored for scientific research. This allows professionals to use AI for previously blocked tasks like drug discovery, clinical development, research biology, and manufacturing.

To participate, organizations must undergo a vetting process covering research credentials, security standards, and ethical oversight. Once verified, teams can apply for two types of grants. Standard Use grants, renewed annually, cover general daily workloads across basic science, quality assurance, and regulatory affairs. High-risk Use grants are project-specific add-ons that must be renewed every six months. These high-risk grants completely remove the safeguards that block life sciences requests. Currently, high-risk grants are available for Claude Opus 5 and Claude Sonnet 5, while high-risk access for Claude Mythos remains restricted to a small set of entities vetted by the US government.

To mitigate risks like insider threats, agent misuse, and access compromise, Anthropic is shifting from real-time blocking to offline monitoring. This approach tracks usage patterns against the applicant's specified use cases, requiring a 30-day data retention period. The data is compartmentalized and excluded from model training. Currently, the LSVP is available via Anthropic's first-party console for API usage, as well as through Claude for Enterprise and Team plans. It is not yet available for individual plans, third-party platforms, or Business Associate Agreement (BAA) enabled organizations handling protected health information. Early adopters of the program include Xaira, Edison, and Manifold Bio.