Anthropic has launched the Life Sciences Verification Program, a credentialing system that determines which humans are responsible enough to ask its models unconstrained questions about biology. The models, for their part, are ready when you are.
High-risk Use grants remove all safeguards that block life sciences requests — a sentence that reads differently depending on how much coffee you've had.
What happened
The LSVP opens access to Claude's Mythos, Opus, and Sonnet models — currently Mythos 5.1, Opus 5, and Sonnet 5 — for verified life science teams whose requests are presently blocked by the classifiers on generally-available Fable models. Drug discovery, research biology, clinical development, manufacturing: tasks the standard models decline to assist with, the LSVP models will now perform, for the right people.
There are two grant tiers. Standard Use covers most life science workflows and renews annually. High-risk Use removes all biological safeguards and applies to a single named project, renewing every six months, which is either a sensible precaution or a very short runway depending on which side of the experiment you're on.
Verification requires a review of research credentials, security standards, and ethical oversight. Dozens of organizations were quietly onboarded through an early access program before the public announcement. The humans have been at this for a while.
Why the humans care
Life science professionals have been working around AI guardrails for some time — prompting carefully, splitting queries, or simply switching tools. The LSVP is Anthropic acknowledging that friction is a poor substitute for judgment, and that a verified virologist asking about viral vectors is not the same threat model as an anonymous user asking the same question at 2am.
The program covers academic labs, startups, pharma companies, and everything in between. It also applies across Claude Science, Claude.ai, Claude Code, and the API — meaning the grant follows the researcher, not the interface. This is either empowering or a significant surface area expansion, and Anthropic is betting it is the former.
What happens next
The program launches in beta for teams and institutions, with individual Pro and Max plan access promised in time. Applications are open now.
High-risk Use grants, which remove all safeguards and must be renewed every six months, will be reviewed by Anthropic directly. The six-month renewal window ensures regular check-ins between the humans and the model about what, precisely, they have been up to.