Monday, 2 November 2026 Tech news and buying guides for the US, UK, Canada and Australia AboutContactRSS
RoxMag
AI

Anthropic Taps Claude to Speed Up Drug Discovery Work

Anthropic is widening Claude's role in pharmaceutical research, with partners reporting faster molecule design, lab testing and document review.

Anthropic Taps Claude to Speed Up Drug Discovery Work

Anthropic is broadening the part its Claude models play in medicine-making, teaming up with biotechnology firms on work meant to shorten the path to new treatments, according to TechRadar. The company described the effort in a new blog post, saying it is studying both scientific and administrative hurdles that slow drug development, from designing molecules to getting ready for clinical trials.

The projects cover possible therapies for cancer, food allergies and uncommon muscle conditions, though most of the candidates are still experimental. Anthropic argues that joining real drug programmes is the surest route to spotting where its models fall short.

At Generate Biomedicines, a team of six scientists turned to Claude Code to rework a piece of CAR T-cell therapy, a technique usually aimed at cancers of the blood. The goal was to free up space for genetic instructions that might let the therapy attack solid tumours. The rebuilt component came out about 70% shorter, and lab work showed immune cells carrying roughly ten times as many receptors. The altered cells also killed cancer cells more effectively and stayed active longer, the findings indicated.

Divergence Labs leaned on Claude to hunt for antibody candidates against IgE, an immune molecule, for a possible food-allergy treatment. In a single month the system produced around 35,000 molecular designs, covering both jobs the proposed therapy needs. Early tests turned up three designs that bind IgE and 20 that pull it away from immune cells, with three of those 20 appearing to outperform the standard drug at that task.

BigHat Biosciences paired Claude with automated lab gear to refine two antibody fragments under consideration for future medicines. Researchers built and screened about 1,010 variants across ten weeks, checking how tightly they bind, how well they hold together and whether they clump. The best cancer-targeting variant attached to its target over ten times more firmly and stayed stable when heated to a level 18 degrees Celsius beyond what the original could withstand. The second fragment clumped more than 100 times more slowly than its predecessor.

Enveda used Claude on manufacturing paperwork that had eaten as much as 75% of some chemists' hours. Over a 1,061-page review, the system caught all 11 problems human reviewers had already flagged and found seven more, cutting manual checking by 89% and taking the job from over 25 hours to under three.

Anthropic has also worked with Bristol Myers Squibb on judging clinical study plans and their odds of success. Michael Pourdehnad, a senior vice president at the drugmaker, said one prompt yielded a solid first-pass analysis, helping weigh options that would otherwise have taken days. The firms plan to go further, looking at Claude's role in early clinical development choices.

Based on reporting by TechRadar. RoxMag adds plain-English context and checks every figure against that source.

anthropicclaudedrug discoverybiotechai research

More in AI