📰 Key Takeaways

Anthropic confirmed to TechCrunch that it operates a wet lab in the Bay Area where its AI models can carry out physical biology experiments. The news echoes recent comments from CEO Dario Amodei, who has said he believes AI could cure most major diseases within the next 5 to 10 years. Anthropic’s head of life sciences, Eric Kauderer-Abrams, told Reuters that when it comes to biology research, “the ultimate test is still going to be work in the real lab, and that’s not going to change anytime soon,” confirming the company is “definitely doing that” right now. He explained that the lab operates like a typical biotech lab, with some research run by Anthropic itself and some done in partnership with outside collaborators. The move isn’t exactly a surprise — Anthropic acquired the stealth-mode AI biotech startup Coefficient Bio back in April. The company hasn’t disclosed what specific research the lab is working on, but it’s stressed that the focus is on fundamental biology rather than drug development, seemingly to avoid looking like it’s competing with its many pharma customers and partners — including its recently announced joint drug development deal with Novo Nordisk. Anthropic has previously taken heat for rolling out products seen as competing with its own customers. The same week, Anthropic also launched a “Life Sciences Verification Program” giving vetted biology researchers access to its most capable models, and published research on accelerating protein design and improving biomolecular modeling. Meanwhile, Anthropic researcher Jacob Coxon recently resigned, warning that “the people building AI genuinely believe this technology could destroy humanity by the end of the century,” and the company’s own head of alignment research has said there’s more than a 10% chance AI causes human extinction within the next decade. Amodei posted over the weekend calling on the industry to slow down and build self-regulation, and has repeatedly named bioterrorism as one of AI’s biggest risks. Investor Chamath Palihapitiya half-jokingly weighed in on the contradiction on X: “the same people screaming ‘we’re all going to die’ and ‘please regulate me’ are building wet labs in San Francisco. i would not recommend this.”

💬 JudyAI Lab Take

Anthropic confirmed to TechCrunch that it runs a wet lab in the Bay Area, letting its AI models carry out physical biology experiments. This one’s worth paying attention to, because it moves the “AI can cure disease” narrative from an executive’s optimistic soundbite into something concrete and operational.

For AI builders, the lesson here is that validating model capability still can’t escape real-world feedback loops. Anthropic’s head of life sciences said it plainly: for biology research, “the ultimate test is still going to be work in the real lab, and that’s not going to change anytime soon.” In other words, even the most capable model still needs a real validation mechanism once it hits a domain that requires physical-world confirmation — you can’t just judge it by how confident the output sounds. The company’s deliberate choice to focus on fundamental biology rather than drug development, so as not to compete with its pharma customers, also shows how AI companies have to balance technical ambition against business partnerships.

Something worth sitting with next time you evaluate any AI capability claim: does this conclusion have a real-world validation loop behind it, or is it just resting on how polished the model’s output sounds?

📅 Source Info

🔗 Further Reading