Anthropic wants to make its own medicine

On this occasion “In Brief: AI for Science” earlier this week, Anthropic declared Claude Sciencenew”AI workbench for scientists“That pulls disparate tools and datasets into one place, and creates statistics and visualizations.” Anthropic, which already dominates the industry with its well-known writing tools and powerful AI models, planned the launch around what it says is AI’s ability to “greatly accelerate the pace of scientific discovery and development of healthcare,” and has already signed up Clauderma’s customer list.

Anthropic went further, saying it would develop its own drug. Director of life sciences Eric Kauderer-Abrams he said the company will focus on finding treatments for “neglected” diseases.

AI companies have been keen to tap scientific and pharma customers – OpenAI, Amazon, Googleand others have their own life science tools and platforms. But Anthropic’s move is one of the biggest AI company’s direct attempts to create medicine itself. It puts it in an unusual position to sell software to other, potential drug-makers. Anthropic joins a growing competition that includes AI-first drug companies like Insilico, Google DeepMind spinout Isomorphic Labs, biotech startups, and Big Pharma companies building or buying their own AI tools.

Anthropic has provided little information on what it hopes to achieve at the drug site. At the event, Kauderer-Abrams did not say what the company would do if it found people hoping to use drugs. Anthropic didn’t answer SeasideRequests for comments require more information, including which diseases to pursue first and whether to partner with other companies for lab work, animal testing, clinical trials, or manufacturing.

Said the experts Seaside that the uncertainty surrounding Anthropic’s plans reflects greater uncertainty in the development of AI medicine. “Finding an AI drug” can mean many things. “It’s a broad term,” explains Namshik Han, a professor at the University of Cambridge and founder of AI biotech startup CardiaTec. AI is used in “every phase of drug discovery,” he said, from discovering new drugs and improving them to supporting research, data analysis, clinical trials, and even manufacturing. Every major pharmaceutical company will be using AI in some way, he said. Matthew Todd, a professor of drug research at University College London, said that AI has already begun to be used in drug research, calling it “an amazing term” because of its wide range of uses.

AI is undoubtedly revolutionizing drug development. Han showed many experiments of pharma giants such as AstraZeneca, Novo Nordisk, and GSK, and said that AI can already help create ideas for potential drugs, such as identifying new molecules that can interact with body parts like receptors on cells that are already known to be involved in certain diseases or that target existing drugs. Todd said it is very helpful in speeding up research and helping to “method test” new drug ideas. Considering Anthropic’s work on frontier species, the company will probably use AI output to explore more possibilities of medicine and ecology and help researchers to make connections that would be difficult or slow to find otherwise, implying new drug concepts, identifying target diseases, or finding new uses for existing drugs.

But it’s still a long way from AI-powered medicine reaching patients. Todd said the field is “far away” from AI-engineered drugs being approved by regulators for human use. He added that the process of obtaining drugs does not run automatically, and is entered and monitored by people at all times. Todd and Han both noted the lack of publicly available, high-quality experimental data, such as how different drugs behave in the body, can limit the efforts of drug development, emphasizing that even in well-trained areas of biology there are large gaps in our understanding of how things work.

AI is not set up to fix many of the slowest parts of drug discovery. Frank von Delft, professor of biological biology at the University of Oxford and head of protein crystallography at the Oxford Center for Medicines Discovery, said that people are right to be excited about the advancement of AI models, but “we have not come close to creating unnecessary experiments.” These agents still need to be tested in the real world to determine if they are effective, toxic, and have useful properties that allow them to be prepared, stored, and administered as medicine. All of this requires skilled workers, a lot of money, and time, especially for clinical work in humans – a time when most promising drugs fail. If Anthropic wants to develop a drug, von Delft said, “they have to spend a lot of money on testing.”

It’s possible Anthropic is ready to try. In the last year, the company has been hiring biologists and building their own wet labsand in writing it has several live programs recruiting for life science positions. Han said Anthropic is “recruiting actively” in the area, adding that several of his fellow students have been approached by the company. Without naming names, Han said he thinks Anthropic has hired a handful of people from Big Pharma and top academic institutions.

With all these complications, any disease that Anthropic chooses, any payment is far away – at least, the better part of a decade, given. how long it often takes a new drug to pass clinical trials. “There’s always a very long time” with experimental drugs, Todd said. It takes time to show experimentally that something is safe. No AI-engineered drug has yet made it through clinical trials and FDA approval to reach the market. Some AI-generated followers have it he entered medical trials, but it is difficult to know how much AI helped, where it was used, or whether the candidates outperformed conventional drugs. AI can speed up the research process, but drugs still need to present themselves in the traditional way: slow, systematic experiments that take place in the real world.