Anthropic is moving deeper into life sciences. Alongside the launch of Claude Science, the company says it plans to work on drugs of its own, a step that puts a major frontier AI company closer to the long, expensive and uncertain process of turning scientific ideas into medicines.
The ambition is clear. The timeline is not. Experts cited in the source article say AI is already useful across drug discovery, but they also warn that the hardest parts of medicine still happen outside the model: in experiments, testing, clinical work and the slow process of proving safety.
What Anthropic Announced
At the event “The Briefing: AI for Science,” Anthropic announced Claude Science, described as an “AI workbench for scientists.” The tool is meant to bring fragmented tools and datasets into one environment and can also generate figures and visuals.
Anthropic framed the launch around AI’s potential to “dramatically accelerate the pace of scientific discovery and the development of healthcare interventions.” The company also pointed to biotech and pharma customers already using Claude.
The more striking part of the announcement was not just software. Anthropic said it would develop drugs itself. Eric Kauderer-Abrams, the company’s head of life sciences, said Anthropic will focus on discovering treatments for “neglected” diseases.
That makes the move different from simply selling AI tools to researchers. Anthropic is entering a space where it may provide software to drugmakers while also pursuing its own drug development work. The source article describes that as an unusual position, because those customers could also become competitors.
Why AI Drug Discovery Is Hard To Define
One reason Anthropic’s plans are difficult to evaluate is that “AI drug discovery” can mean many different things. Namshik Han, a professor at the University of Cambridge and cofounder of AI biotech startup CardiaTec, described it as “a really broad term.”
Han said AI is applied at “every single stage of drug discovery.” That can include finding new compounds, improving them, supporting research, analyzing data, helping with clinical trials and even manufacturing.
Matthew Todd, a professor of drug discovery at University College London, made a similar point. He called AI drug discovery a “catchall phrase,” because the technology can be used in so many parts of research and development.
That breadth matters. A company can use AI to organize scientific work, suggest molecules, examine data or help researchers test ideas before moving into the lab. Those are all meaningful, but they are not the same as delivering a medicine to patients.
Where AI Can Help
The source article makes clear that AI is already changing drug development. Han pointed to initiatives by pharma giants including AstraZeneca, Novo Nordisk and GSK. He said AI can help generate possible drug ideas, including by suggesting molecules that might interact with parts of the body such as cell receptors already known to be involved with a disease or targeted by existing drugs.
Todd said AI can be valuable for speeding up research and helping “road test” new drug ideas. Given Anthropic’s work on frontier models, the source article says the company would presumably use generative AI to search across chemical and biological possibilities and help researchers find connections that would otherwise be difficult or slow to identify.
That could mean several kinds of work:
- suggesting new drug ideas;
- identifying new disease targets;
- finding new uses for existing drugs;
- helping researchers analyze fragmented scientific information;
- supporting parts of the development process that depend on data and comparison.
These uses are important because drug discovery begins with a large search problem. Researchers must decide what biological target matters, what molecule might affect it and whether that idea deserves further testing. AI can make parts of that search faster and broader.
What AI Still Cannot Replace
The difficult part is that a promising idea is not a medicine. Todd said the field is “a long way off” from an AI-designed drug being approved by regulators for human use. He also said the process would not run autonomously, because human input and supervision are needed throughout.
Data is another constraint. Todd and Han both noted the lack of publicly available, high-quality experimental data, including data about how chemicals behave in the body. They also stressed that even well-studied areas of biology still have major gaps in understanding.
Frank von Delft, a professor of structural chemical biology at the University of Oxford and head of protein crystallography at the Oxford Centre for Medicines Discovery, said AI models “haven’t yet come close to making experiments unnecessary.”
That point is central to Anthropic’s drug ambitions. Drug candidates still need to be tested in the real world for efficacy, toxicity and practical properties that determine whether they can be prepared, stored and delivered safely as medicines. Those steps require skilled workers, substantial money and time.
Clinical work in humans is especially difficult, and the source article notes that many promising drug candidates fail at that stage. If Anthropic wants to develop a drug, von Delft said it is “going to have to spend a lot on experiments.”
The Long Road Ahead
Anthropic has shared few specific details about what it wants to accomplish in drug development. Kauderer-Abrams did not say what the company would do if it found promising drug candidates. The company also did not respond to The Verge’s requests for more details, including which diseases it would target first and whether it would partner with other companies for lab work, animal testing, clinical trials or manufacturing.
There are signs Anthropic is preparing for more life sciences work. The source article says that in the last year the company has been actively hiring biologists and building its own wet labs, and that it had several live applications for life sciences roles. Han said Anthropic has been “actively recruiting” in the area and said several of his academic colleagues had been approached by the company.
Still, any payoff is likely far away. The source article says that, whatever disease Anthropic chooses, the result is likely a long way off, at the very least the better part of a decade, because of how long new drugs typically take to move through clinical trials.
Todd said there is “always a big lag time” with testing medicine and added, “It takes time to show experimentally that something’s safe.” No AI-designed drug has yet made it through clinical trials and FDA approval to reach market. Some AI-developed candidates have entered clinical trials, but it remains hard to know how much AI contributed, where it was used or whether those candidates outperform conventional drugs.
Anthropic’s move is a notable signal for the AI drug discovery race. It shows that frontier AI companies may want to do more than provide tools. But the source article’s experts point to the same conclusion from several angles: AI can speed up parts of the search, while medicines still have to prove themselves through slow, methodical experiments in the real world.