A Special Path Just for Life Sciences Pros: Anthropic Launches the Life Sciences Verification Program

Hi everyone, it's me! Today I want to walk you through an announcement that sits right at the intersection of AI and biology, so let's dig in together.

Anthropic News anthropic.com

What was announced?

Anthropic's News published the launch of the Life Sciences Verification Program (LSVP). It's a new access program that gives verified life science professionals access to the Mythos, Opus, and Sonnet models with a more permissive set of safeguards for biology-related work. It's aimed at teams ranging from academic labs to pharmaceutical companies.

The story so far

Until now, Claude models relied on real-time blocking to stop high-risk biology-related requests. That approach helps prevent misuse, but it also meant legitimate researchers doing drug discovery or clinical development work could get blocked outright when asking for information they genuinely needed. Anthropic's threat reporting has shown increasingly sophisticated abuse attempts on its platform, and biology is a field where it's especially hard to tell legitimate research apart from harmful intent.

What changes

Verified organizations can now apply for two kinds of access, depending on what they need:

• Standard Use: covers most life science work, spanning basic science, R&D, supply chain and manufacturing, clinical development, quality assurance, and regulatory affairs. It renews annually and can be extended across a whole team.
• High-risk Use: an add-on for work that's blocked under Standard Use, scoped to a single research project. One example given is characterizing how a specific viral vector is recognized by human immune pathways. It requires renewal every six months.

In short, verified researchers can now get answers to advanced biology questions that used to be blocked outright.

Dive Deep

The core shift here is moving safety from real-time blocking to continuous, offline monitoring. The program addresses three main threat models:

• Credential theft or account takeover leading to unauthorized access
• Insider threats from within an organization
• Agentic misuse, where an agent takes unintended risky actions during a long-running task

Usage is checked against the intended use case declared at application time, and if a suspicious pattern is flagged, it's reported to the organization's admin, who must respond within an agreed-upon window. Data kept for this monitoring is retained for 30 days, strictly compartmentalized, never used for model training, and not accessible to members of Anthropic's own life sciences research team.

The program is available across Claude Science, Claude.ai enterprise plans, Claude Code, and the API (individual Pro and Max plans aren't supported yet, but are planned). Organizations apply through a dedicated form, and verification checks research credentials, security standards, and ethical research oversight. Companies like Xaira, Edison, and Manifold Bio have already shared supportive comments about the program. Anthropic expects hundreds of organizations to sign up in the first few weeks, with more updates on new products, research collaborations, and program improvements coming in the following months.

Wrap-up

• Anthropic launched the Life Sciences Verification Program for verified life science professionals
• Verified organizations can apply for Standard Use (annual renewal, broad workflows) or High-risk Use (six-month renewal, single project)
• Safety now relies on continuous monitoring and insider-threat safeguards instead of real-time blocking
• Available via Claude Science, Claude.ai enterprise, Claude Code, and the API, with individual plans coming later

If you're a researcher or company in drug discovery or clinical research who's been running into safety blocks, this program is worth checking out!