Anthropic plans move from AI science tools into drug discovery
Anthropic says it wants to develop treatments of its own, even as experts warn AI-designed drugs still face years of lab and clinical testing.
By Maya Lindqvist · Senior Technology Correspondent
3 min read
Anthropic is preparing to do more than sell AI software to scientists: it also wants to work on medicines of its own. The plan puts a major AI model developer closer to the expensive, slow drug pipeline at a time when no AI-designed drug has yet reached the market with FDA approval, according to The Verge.
The company introduced Claude Science this week at its “The Briefing: AI for Science” event, describing it as an AI workbench meant to bring scattered research tools and datasets into one place. Anthropic said the product can also help generate scientific figures and other visuals, and it presented the launch as part of a push to speed scientific research and health care development.
Anthropic’s head of life sciences, Eric Kauderer-Abrams, told CNBC that the company also plans to discover its own drugs, with an initial focus on treatments for “neglected” diseases. The Verge reported that Anthropic has not offered many specifics about how that effort would work, including which diseases it would target first or whether it would use outside partners for lab work, animal studies, clinical trials or manufacturing.
A crowded AI drug field
AI companies have been building tools for pharmaceutical and life sciences customers, including OpenAI, Amazon and Google, according to The Verge. Anthropic’s plan goes further by placing the company in the role of a potential drug developer while it also sells software to drugmakers that could become competitors.
The Verge reported that Anthropic would enter a field that already includes AI-focused drug companies such as Insilico, Isomorphic Labs, biotech startups and large pharmaceutical companies developing or buying AI tools. Experts told The Verge that the phrase “AI drug discovery” covers many different uses, from finding molecules and analyzing data to supporting clinical trials and manufacturing.
Namshik Han, a University of Cambridge professor and cofounder of CardiaTec, told The Verge that AI is used across drug discovery and that major drug companies are likely to use it in some form. Matthew Todd, a University College London professor of drug discovery, told The Verge that AI can speed research and test drug ideas, but he said the field remains far from getting an AI-designed drug approved for human use.
Experiments remain the bottleneck
Experts told The Verge that AI may help researchers search chemical and biological possibilities, suggest new drug ideas, identify disease targets or find new uses for existing medicines. They also warned that drug candidates still have to prove safety, effectiveness and practical medical properties through experiments.
Frank von Delft, an Oxford professor of structural chemical biology and head of protein crystallography at the Oxford Centre for Medicines Discovery, told The Verge that current AI models have not removed the need for experiments. He said any Anthropic drug project would require substantial spending on real-world testing.
The Verge reported that Anthropic has been hiring biologists, building wet labs and listing life sciences jobs. Han told The Verge that Anthropic has been recruiting in the field and said some academic colleagues had been approached by the company.
Even if Anthropic identifies a promising candidate, experts told The Verge that clinical development could take years. Todd said medicine testing carries a long delay because safety must be shown experimentally, and The Verge noted that some AI-developed drugs have entered clinical trials, though it remains difficult to judge how much AI contributed or whether those candidates perform better than conventional drugs.
This story draws on original reporting from The Verge.