Danny Weber
Claude Science is pitched as a workspace for researchers, while Anthropic says it wants to work on treatments for rare and neglected diseases.
Anthropic has announced the launch of Claude Science, a platform for researchers, while also saying it wants to get directly involved in drug development. The company plans to focus on rare and “neglected” diseases, which often fall outside the priorities of major pharmaceutical companies because the commercial upside is limited. For now, the focus is mainly on early drug discovery and preclinical research.
Claude Science is presented as a single workspace for scientists. The platform is meant to bring scattered tools, datasets, analysis features, charting and scientific support into one place. During the presentation, Anthropic cited an example from UCSF: with Claude Science, a researcher reportedly found viral contamination in an experiment within minutes, something the team had missed for about a year.
The company also claims the system can analyze 100 rare genetic diseases in under an hour and identify 32 promising directions for further computer-based screening. Unlike many AI companies that limit themselves to supplying tools for the pharmaceutical market, Anthropic says it wants to participate directly in drug development. Still, the company has not yet named its first disease candidates or explained whether it will carry projects forward itself or bring in partners for animal studies, clinical trials and manufacturing.
Interest in AI across pharma is rising quickly: OpenAI, Google, Amazon and other major players already offer tools for biotechnology and medicine, while Google DeepMind, through Isomorphic Labs, and companies such as Insilico Medicine are pushing their own drug-development efforts. Traditional pharma giants, including AstraZeneca, Novo Nordisk and GSK, are also actively using AI to find molecules, analyze data, design compounds and optimize R&D workflows.
Experts, however, warn that AI is still an accelerator, not a replacement for full drug development. Researchers from Cambridge, UCL and Oxford note that promising candidates still need toxicology, safety checks, efficacy assessments, animal testing and clinical studies in humans. Novartis CEO Vas Narasimhan believes new AI tools could cut the average drug-development cycle from about 12 years to 7–8 years and potentially lift project success rates from 8% to 16%, but biology and regulation cannot be skipped.
Anthropic is already expanding its life-sciences team, building its own wet lab and hiring specialists from biology, pharma and research institutes. Even if Claude Science really does speed up the search for promising directions, clinical results are still a long way off: drug development remains expensive, slow and tightly regulated.
© RusPhotoBank