Anthropic operates, according to research by the news agency Reuters, its own biology lab in the San Francisco Bay Area, where researchers conduct physical experiments. Life Sciences Chief Eric Kauderer-Abrams publicly confirmed the operation for the first time after the company had not communicated about the facility for months. Clinical trials are currently excluded by Anthropic, according to its own statements.
Coefficient Bio Acquisition Lays Foundation for the Lab
In April, Anthropic acquired the previously little-known biotech company Coefficient Bio for around 400 million dollars in stock. The purchase laid the foundation for the company’s own life sciences branch. At the end of June, a working environment called Claude Science followed, which consolidates databases and tools for genomics, proteomics, and drug research.
Life Sciences Chief Kauderer-Abrams had previously co-founded the diagnostics company Detect and brought a COVID-19 home test approved by the U.S. Food and Drug Administration to market. He justifies the physical lab work by stating that biological hypotheses can ultimately only be confirmed in real experiments. A purely digital model is not sufficient for this. Therefore, Anthropic categorizes the work as basic biological research, not its own drug development.
For competitive reasons, the company has set a clear boundary: it will not conduct its own clinical trials for the time being but will focus on questions that the established pharmaceutical industry has not yet addressed. Additionally, Anthropic has launched a verification program through which vetted external researchers gain access to the most powerful in-house models. The company did not want to disclose details about ongoing projects in the lab itself to reporters.
Pharmaceutical Companies Test Claude in Their Own Research Processes
In addition to its own lab work, Anthropic supplies several pharmaceutical companies with AI tools. Genentech and Bristol Myers Squibb have been customers for some time. Since September 16, Novo Nordisk has joined as another major manufacturer. The Danish company is testing Claude Science in selected research processes and additionally uses Anthropic’s models in its own software development.
Novo Nordisk CEO Mike Doustdar hopes for increased productivity in research and development from this. Anthropic CEO Dario Amodei sees such partnerships as an opportunity to significantly shorten research timelines. Google DeepMind and OpenAI are also simultaneously building their own offerings for biomedicine. The three major AI providers are increasingly competing for pharmaceutical customers as a growth area beyond chatbots and coding tools.
Part of Anthropic’s lab work also concerns automation: the company is testing whether Claude can guide robotic systems in such a way that experiments run with minimal human presence. However, a spokesperson emphasized that human involvement remains mandatory for safety reasons. The technical foundation for this was already laid by Anthropic in August with an open standard for AI-controlled lab robots, which is being tested by Genentech and the quantum computing company QuEra, among others. Fixed speed and angle limits are intended to prevent the machines from acting uncontrollably.
It remains to be seen whether the previously quietly operated facility will become an independent pharmaceutical business or whether Anthropic will remain in the role of tool supplier for Novo Nordisk, Genentech, and Bristol Myers Squibb. The confirmation also comes at a time when security researchers at the company have publicly expressed concerns about Anthropic’s course. How robust the promised human oversight in the lab’s daily operations actually is cannot yet be verified from the outside.
