Anthropic Aims to Create Its Own Pharmaceuticals
At the recent event “The Briefing: AI for Science,” Anthropic unveiled Claude Science, an innovative “AI workbench for scientists” designed to consolidate various tools and datasets into a single platform, enabling the generation of figures and visuals. With a strong foothold in the industry thanks to its popular coding tools and advanced AI models, Anthropic emphasized the launch as a means to “dramatically accelerate the pace of scientific discovery and the development of healthcare interventions,” highlighting a growing list of biotech and pharmaceutical clients already utilizing Claude.
In a bold move, Anthropic also announced plans to develop its own drugs. Eric Kauderer-Abrams, head of life sciences, stated that the company aims to focus on treatments for “neglected” diseases. The interest from AI companies in the science and pharma sectors is palpable, with major players like OpenAI, Amazon, and Google also offering their own life sciences tools. However, Anthropic’s approach stands out as one of the most direct efforts by a leading AI firm to engage in drug development, placing it in a unique position of selling software to potential competitors in the pharmaceutical space. This places Anthropic in a competitive landscape alongside AI-driven drug companies like Insilico and Isomorphic Labs, as well as established biotech firms and Big Pharma investing in AI technologies.
Details on Anthropic’s drug development ambitions remain sparse. During the event, Kauderer-Abrams did not clarify what steps the company would take if promising drug candidates were identified. Requests for further information, including targeted diseases and potential partnerships for lab work, animal testing, clinical trials, or manufacturing, went unanswered.
Experts have noted that the ambiguity surrounding Anthropic’s plans reflects a broader uncertainty in the AI drug discovery landscape. The term “AI drug discovery” encompasses a wide range of applications, as explained by Namshik Han, a professor at the University of Cambridge and cofounder of an AI biotech startup. He pointed out that AI is utilized at every stage of drug discovery, from identifying and refining new compounds to aiding in research, data analysis, clinical trials, and manufacturing. Matthew Todd, a professor of drug discovery at University College London, echoed this sentiment, describing AI as a “catchall phrase” due to its diverse applications in the field.
AI is undeniably transforming drug development. Han highlighted initiatives from major pharmaceutical companies like AstraZeneca and GSK, noting that AI can assist in generating potential drug ideas by suggesting new molecules that interact with known disease-related cell receptors. Todd emphasized the technology’s role in accelerating research and “road testing” new drug concepts. Given Anthropic’s expertise in frontier models, it is likely that the company will leverage generative AI to explore vast chemical and biological possibilities, potentially uncovering new drug ideas, identifying disease targets, or discovering new applications for existing medications.
However, the journey from AI-generated concepts to approved drugs is still lengthy. Todd remarked that the industry is “a long way off” from seeing an AI-designed drug approved for human use, emphasizing that human oversight will be essential throughout the drug discovery process. Both Todd and Han pointed out that the lack of publicly available, high-quality experimental data on chemical behavior in the body could hinder drug development, highlighting significant gaps in our understanding of biological processes.
AI is not equipped to address many of the slowest aspects of drug discovery. Frank von Delft, a professor at the University of Oxford, cautioned that while advancements in AI models are exciting, they have not yet eliminated the need for real-world experiments. Drug candidates must still undergo rigorous testing for efficacy, toxicity, and practical properties, which requires skilled personnel, substantial funding, and time—especially during human clinical trials, where many promising candidates fail. If Anthropic intends to develop a drug, von Delft noted, it will need to invest heavily in experimental work.
Anthropic appears to be preparing for this challenge. Over the past year, the company has been actively recruiting biologists and establishing its own wet labs, with several job openings in life sciences currently available. Han mentioned that Anthropic has been “actively recruiting” in this field, with some of his academic colleagues reportedly approached by the company. He believes Anthropic has successfully attracted talent from both Big Pharma and prestigious academic institutions.
Given the complexities involved, any potential breakthroughs from Anthropic are likely years away—typically a decade or more, considering the lengthy clinical trial process for new drugs. Todd noted that there is “always a big lag time” in medicine testing, as it takes time to demonstrate safety through experimentation. While some AI-developed candidates have entered clinical trials, it remains unclear how much AI contributed to their development or whether they outperform traditional drugs. AI can expedite certain aspects of the search, but ultimately, drugs must validate their efficacy through traditional, methodical experiments conducted in real-world settings.
