Anthropic Launches a Verification Program Giving Life Sciences Researchers Fewer-Interruption Access to Claude
Anthropic launched the Life Sciences Verification Program (LSVP), giving vetted life sciences professionals access to its Mythos, Opus, and Sonnet models under safeguards tailored to biology research, replacing real-time blocking with offline monitoring to reduce interruptions to legitimate work.
Details
- Who it’s for: academic laboratories, startups, pharmaceutical companies, and research institutions working in drug discovery, research biology, clinical development, manufacturing, supply chain, quality assurance, regulatory affairs, and investing
- Verification process: applicants are reviewed on research credentials, security standards, and ethical research oversight, and must describe intended use cases at a high level — similar to a job posting — without sensitive information or IP
- Two access tiers: Standard Use covers most life science work, is renewed annually, and is available to entire teams; High-risk Use is a project-specific add-on requiring additional vetting, renewed every six months
- Monitoring approach: usage is checked against stated use cases with 30-day data retention on flagged activity; data is compartmentalized and excluded from model training or Anthropic’s own life sciences research
- Partner organizations: Xaira, Edison, and Manifold Bio are quoted supporting the program
- Availability: accessible now via the API, Claude Science, Claude for Enterprise, and Team plans — not yet available on individual plans or third-party platforms, and organizations with a Business Associate Agreement are excluded during the beta
What happened next
The program addresses a specific tension in AI safety for dual-use biology research — safeguards designed to block bioweapon-relevant queries can also slow down legitimate drug discovery work — by shifting from blocking suspicious requests outright to reviewing flagged activity after the fact for verified, credentialed researchers.