Anthropic is launching the Life Sciences Verification Program (LSVP) in beta. It gives verified life science professionals access to Anthropic's Mythos, Opus, and Sonnet models with safeguards that are more permissive for biology-related work. Dozens of organizations have already joined through early access, and Anthropic is now accepting waitlist applications from the wider life science community. The beta is initially open to teams and institutions. Anthropic plans to extend it to individual Pro and Max plans over time.
The program is meant to unlock tasks that Anthropic's generally available models currently block, such as drug discovery, research biology, clinical development, and manufacturing. It is intended for all kinds of teams, from academic labs to startups and pharma companies.
Verification and access types
Every applicant goes through verification, which reviews their research credentials, security standards, and ethical research oversight. Verified teams can then apply for one of two grant types, Standard Use or High-risk Use, depending on what access they need. Grants work across all of Anthropic's product surfaces, including Claude Science, Claude.ai, Claude Code, and the API.
Standard Use grants cover most life science work, including the bulk of biology research and development workflows. They can be extended to whole teams for varied daily workloads, and they renew annually. They provide access to Anthropic's Mythos, Opus, and Sonnet models, using refined classifiers that allow more science tasks than the generally available models do. Today they apply to Mythos 5.1, Opus 5, and Sonnet 5, and they will cover future models as those launch. They are built to support the full range of life science activity: basic science, R&D, supply chain and manufacturing, clinical development, quality assurance, regulatory affairs, investing and diligence, and more.
Anthropic expects Standard Use to meet most access needs. Some work has a higher potential for misuse, though, and needs extra vetting.
High-risk Use is an add-on grant for teams whose work falls into areas that Standard Use blocks. It removes all safeguards that block life sciences requests. It applies to a single research project rather than a whole team, and it must be renewed every six months. A typical researcher doing dual-use work would hold one Standard Use grant for varied daily activity, plus one or more High-risk Use grants tied to specific projects. One example is characterizing how a particular family of viral vectors is recognized by human immune pathways.
High-risk grants for Claude Opus 5 and Claude Sonnet 5 are available now. Anthropic is working with the US government to widen availability of high-risk grants for Claude Mythos. At launch they stay limited to a small set of entities that undergo additional vetting.
All other safeguards, such as cyber classifiers, stay in place under LSVP grants.
Enabling trusted access through shared responsibility
Anthropic's recent threat report shows increasingly sophisticated misuse attempts on its platform, including attempts that could support biological weapons development. In biology, it is often impossible to tell valid work apart from harmful intent. Examples are researching a viral pathogen to develop vaccines versus trying to make a virus more transmissible. The most worrying threat models are therefore ones where legitimate access is diverted or taken over by a bad actor. Insider threats and rogue use have played major roles in significant biosafety incidents and scares. In designing the LSVP's safeguards, Anthropic targeted three threat models:
- Access compromise: Malware or account takeover that diverts access to a bad actor.
- Insider threats: Rogue or coerced employees who take malicious action themselves or divert their access to a bad actor.
- Agent misuse: Agents, especially those working in swarms or on long-horizon tasks, taking unintended dangerous actions.
To defend against these threats, and working closely with enterprise CISOs, Anthropic built the safeguards around shared responsibility. Usage is monitored against the intended use case for the model access. Because Anthropic vets LSVP organizations for life sciences credibility and oversight, it can let them define for themselves what safe usage means for their teams or projects.
Each entity's access is tied to the use cases named in its grant applications. Anthropic continuously monitors LSVP traffic for usage or patterns outside the stated scope. When unauthorized activity occurs, Anthropic can flag it to organization admins, who act within pre-agreed timeframes to triage and remediate. Use cases should be high-level descriptions of the intended work, like what you would put in a job listing, and should not include sensitive information or IP.
How monitoring works in LSVP
Serious misuse is often split across many requests and sessions so that the pieces look unrelated and escape detection. In the LSVP, Anthropic is moving safeguards away from real-time blocking, which rejects potentially harmful requests as they arrive, and toward offline monitoring, which better reveals misuse across patterns of behavior. This lets legitimate work continue with fewer interruptions, but it means Anthropic must retain data tied to flagged activity for review. For LSVP traffic, Anthropic requires 30 days of data retention so it can monitor effectively.
This data is strictly compartmentalized. It cannot be used for model training, and Anthropic's life sciences research teams cannot access it. For qualifying organizations, Anthropic is also exploring how the LSVP can integrate with its Enterprise Frontier Safeguards (EFS) systems.
What researchers are saying
Xaira Therapeutics says it is making biology more computable by generating biological data at unprecedented scale and building foundation models of cell, protein, and disease biology. It is excited to apply Anthropic's frontier models across its drug discovery engine and believes pairing trusted access with intelligence is the right way to realize AI's promise in biology.
Edison Scientific says its mission is to accelerate science and the discovery and development of new medicines. It is excited to apply Anthropic's most intelligent models to these problems through the program, and it hopes to collaborate further with Anthropic to end disease and improve patients' lives.
Manifold Bio says it is building a massively parallel interface into living systems so powerful AI can create medicines. It looks forward to safely applying frontier intelligence in its engine and welcomes Anthropic's approach of pairing access with accountability.
Applications and availability
Organizations interested in joining can register interest through the program form. Anthropic expects to enroll hundreds of organizations in the first week and to scale up in the coming weeks to serve most of the life science community.
Today the LSVP is available in Anthropic's first-party console for API usage, and in Claude for Enterprise and Team plans. Individual plans are not yet supported, though Anthropic is working to expand access to them. It is also not yet available on third-party platforms.
Because it is a beta, the LSVP is not available for BAA-enabled orgs. Customers with PHI data should use separate non-BAA, non-HIPAA orgs.
In the API and Claude Science, users can switch between grants natively. In Claude.ai and Claude Code, initially only a preselected default grant applies, except when using Claude Code with API authentication. That should suit the vast majority of users, who will only ever need a Standard Use grant. Anthropic will improve support and portability of these features over time.
What comes next
Offering these frontier capabilities is part of Anthropic's broader effort to support the life sciences community in a shared mission to accelerate cures for disease and improve human health. Anthropic will share more about new products, research collaborations, and program improvements in the coming months.
Changelog:
September 30th, 2026. Anthropic clarified that the linked form is an expression of interest, not an application.