Anthropic Life Sciences Verification Program
Anthropic has introduced the the Life Sciences Verification Program (LSVP), a beta initiative providing verified life science professionals with access to Mythos, Opus, and Sonnet models under a refined set of safeguards. This program allows researchers and organizations to perform biology-related tasks—such as drug discovery, clinical development, and manufacturing—that are typically blocked by the safeguards in generally available models.
Access Tiers and Verification
To qualify for the LSVP, applicants must undergo a verification process reviewing their research credentials, ethical research oversight, and security standards. Once verified, organizations can apply for two distinct types of grants:
Standard Use Grants
Standard Use grants are designed for the majority of biology research and development workflows. These grants provide access to Mythos 5.1, Opus 5, and Sonnet 5 (and future models) with classifiers that are more permissive for science tasks than standard models. They are renewed annually and can be extended to entire teams across various domains, including basic science, R&D, supply chain, regulatory affairs, and quality assurance.
High-risk Use Grants
High-risk Use is an add-on grant for specific research projects that carry a higher potential for misuse and would be blocked under Standard Use. These grants remove all safeguards that block life sciences requests. Because of the increased risk, High-risk Use grants are project-specific rather than team-wide and must be renewed every six months.
Currently, High-risk grants are available for Claude Opus 5 and Claude Sonnet 5. Access for Claude Mythos is limited to a small set of entities with additional vetting while Anthropic works with the US government to expand availability.
Security Model and Shared Responsibility
Anthropic's security approach for the LSVP is based on a shared responsibility model to mitigate three primary threat models: access compromise (account takeovers), insider threats (rogue employees), and agent misuse (unintended dangerous actions by AI agents).
To manage these risks, Anthropic monitors usage against the "intended use-case" specified by the organization during the grant application. Organizations define what constitutes safe usage for their teams, and Anthropic monitors traffic to identify patterns outside that stated scope. If unauthorized activity is detected, Anthropic flags the cases to organization administrators for triage and remediation within pre-agreed timeframes.
Monitoring and Data Retention
To detect sophisticated misuse that may be spread across multiple sessions to evade detection, the LSVP shifts from real-time blocking to offline monitoring. This transition allows legitimate scientific work to proceed with fewer interruptions.
To support this offline monitoring, Anthropic requires 30 days of data retention for LSVP traffic. This data is strictly compartmentalized and is not used for model training or accessed by Anthropic's life sciences research teams.
Availability and Integration
The LSVP is currently available in beta for teams and institutions via the first-party console for API usage, Claude for Enterprise, and Team plans. It is not yet available for individual Pro and Max plans or third-party platforms.
Additionally, the program is not available for BAA-enabled organizations; customers handling PHI data must use separate non-BAA organizations with non-HIPAA configurations.
Industry Perspectives
Several life science organizations have expressed support for the program's balance of access and accountability:
Xaira is making biology more computable... We’re excited to put Anthropic's frontier models to work across our drug discovery engine, and we believe pairing trusted access with intelligence is the right way to realize AI’s promise in biology.
Edison’s mission is to accelerate science... With the Life Sciences Verification Program, we are excited to be able to bring Anthropic’s most intelligent models to bear on these problems.
At Manifold Bio, we’re building a massively parallel interface into living systems... We look forward to putting frontier intelligence to work safely in our engine, and we welcome Anthropic’s approach of pairing access with accountability.