Who can apply to the Life Sciences Verification Program?

The Life Sciences Verification Program is open to life sciences organizations and institutions rather than individuals. Anthropic runs it through its first-party Console for API usage and through Claude for Enterprise and Team plans. Individual plans are not supported yet, and the program is not yet available on third-party platforms or for BAA-enabled organizations.

What is the difference between Standard Use and High-risk Use access?

Standard Use covers most biology research, including drug discovery, clinical development and manufacturing, and is renewed once a year for an entire team. High-risk Use removes the life-sciences-specific safeguards for particular research projects and must be renewed every six months. Anthropic grants each tier against the use cases an organization states in its application.

Does Anthropic train Claude on Life Sciences Verification Program traffic?

No. The Life Sciences Verification Program requires 30 days of data retention so that flagged activity can be reviewed, but Anthropic states this data is restricted from model training. The retention supports offline detection of misuse patterns across sessions rather than feeding future Claude models.

What happens if a verified team's work drifts outside its stated use cases?

Organizations name their intended use cases when they apply, and Anthropic flags Life Sciences Verification Program activity that falls outside them. The stated threat models are access compromise, insider threats and agent misuse, so the monitoring looks across sessions offline instead of blocking any single prompt in real time.

Anthropic · 2026-09-17 · major

Life Sciences Verification Program — Anthropic loosens Claude's biology limits

Anthropic's Life Sciences Verification Program gives vetted research organizations Claude Mythos, Opus and Sonnet with classifiers tuned to be more permissive for biology work. Applications are now open to the life sciences community.

Anthropic Life Sciences Verification Program announcement illustration showing a DNA double helix

Vetted life-science teams get a Claude that does not stop short on ordinary biology work.

Quick facts

Maker	Anthropic
Models covered	Claude Mythos, Opus, Sonnet
Tiers	Standard (yearly) / High-risk (6 months)
Where	First-party Console API, Claude for Enterprise, Team
Not supported	Individual plans, third-party platforms, BAA-enabled orgs
Data retention	30 days, excluded from model training
Status	Beta, applications open

What is it?

The Life Sciences Verification Program (LSVP) gives approved research organizations Claude Mythos, Opus and Sonnet with classifiers retuned to be more permissive for science tasks. Anthropic launched it in beta for teams and institutions, and applications are now open to the broader life sciences community. Access comes in two tiers, one renewed yearly and one renewed every six months.

How does it work?

Rather than blocking a prompt as it arrives, LSVP leans on offline monitoring. An organization states its intended use cases when it applies, and Anthropic then reviews traffic across sessions for activity outside those stated objectives. Flagged activity carries a 30-day retention requirement, and that data is restricted from model training.

Why does it matter?

Drug discovery, clinical development and manufacturing work regularly trips safety classifiers written for general use, so researchers lose time rewording prompts or give up on the model. Moving the check from every message to the vetted organization is what makes long biology workflows practical inside Claude for Enterprise and the API.

Who is it for?

life sciences and drug discovery teams

Frequently asked questions

Who can apply to the Life Sciences Verification Program?: The Life Sciences Verification Program is open to life sciences organizations and institutions rather than individuals. Anthropic runs it through its first-party Console for API usage and through Claude for Enterprise and Team plans. Individual plans are not supported yet, and the program is not yet available on third-party platforms or for BAA-enabled organizations.
What is the difference between Standard Use and High-risk Use access?: Standard Use covers most biology research, including drug discovery, clinical development and manufacturing, and is renewed once a year for an entire team. High-risk Use removes the life-sciences-specific safeguards for particular research projects and must be renewed every six months. Anthropic grants each tier against the use cases an organization states in its application.
Does Anthropic train Claude on Life Sciences Verification Program traffic?: No. The Life Sciences Verification Program requires 30 days of data retention so that flagged activity can be reviewed, but Anthropic states this data is restricted from model training. The retention supports offline detection of misuse patterns across sessions rather than feeding future Claude models.
What happens if a verified team's work drifts outside its stated use cases?: Organizations name their intended use cases when they apply, and Anthropic flags Life Sciences Verification Program activity that falls outside them. The stated threat models are access compromise, insider threats and agent misuse, so the monitoring looks across sessions offline instead of blocking any single prompt in real time.

Try it

https://claude.com/form/life-sciences-verification-program