Anthropic Shifts Strategy to Develop Own Drugs for Neglected Diseases

Anthropic, the artificial intelligence startup backed by billions in venture capital, is transitioning from a software provider to a direct participant in the pharmaceutical industry. During The Briefing: AI for Science event, the company revealed plans to develop its own medications, specifically targeting conditions currently overlooked by major commercial laboratories.

Eric Kauderer-Abrams, Anthropic’s head of life sciences, told CNBC that the company intends to “experiment in first person” with the complexities of drug discovery. This strategy aims to bridge the gap between theoretical AI modeling and the practical hurdles of the laboratory. By developing its own pipeline, Anthropic seeks to refine its Claude Science tools through direct application rather than remaining a detached technology vendor.

The initiative focuses on neglected diseases, a category often sidelined by the traditional pharmaceutical industry due to low profit margins or complex pathology. Anthropic suggests its selection of programs will be driven by patient benefit rather than purely commercial market potential, according to a company spokesperson.

Historically, the path from molecule discovery to market is fraught with failure. The industry often refers to the “valley of death,” where promising laboratory candidates fail during human trials. AI has been marketed as a way to shorten this decade-long process, which can cost upwards of $2 billion per successful drug.

However, academic experts caution against overestimating the autonomy of current AI models. Matthew Todd, a professor of drug discovery at University College London, noted that the approval of a drug designed entirely by AI remains a distant prospect. While AI assists in various stages, human oversight remains a regulatory and scientific necessity.

Frank von Delft, a professor of structural biology at the University of Oxford, emphasized that AI cannot yet bypass physical experimentation. Real-world testing is required to verify efficacy and toxicity—factors that determine whether a compound can be safely administered to humans.

Despite these hurdles, Anthropic is actively building the infrastructure for a physical science footprint. Namshik Han, a professor at the University of Cambridge and co-founder of CardiaTec, confirmed that the company has been recruiting biologists and establishing its own laboratory facilities. This indicates a shift toward a hybrid model where silicon-based prediction meets traditional wet-lab validation.