Anthropic to Develop Drugs Using AI, Reports Say

Anthropic launches AI platform for drug development.

AI startup Anthropic has launched a scientific platform called Claude Science and announced plans to develop treatments for advanced diseases, according to The Verge.

The platform was unveiled at The Briefing: AI for Science event. According to the announcement, it integrates disparate tools and datasets, and creates graphs and visualizations.

Anthropic stated that clients in the biotech and pharmaceutical industries are already using the product.

The Verge noted that companies like OpenAI, Amazon, and Google already offer medical tools. However, Anthropic is the first to announce plans to independently develop drugs, beyond just providing software.

The company did not respond to journalists’ questions about specific research directions or types of diseases.

According to the publication, Anthropic has been actively hiring biologists and building its own laboratories over the past year. Namshik Han, a professor at Cambridge University and co-founder of CardiaTec, mentioned that the company approached his academic colleagues with recruitment offers.

He noted that the AI project has also attracted specialists from AstraZeneca, Novo Nordisk, GSK, and other major scientific institutions.

Experts interviewed by The Verge confirmed that artificial intelligence is already used at various stages of drug development, but it does not replace experiments and human labor.

Han listed typical scenarios such as the search and optimization of chemical compounds, in-depth research, data analysis, clinical trials, and production.

Matthew Todd, a professor at University College London, added that “drug development with AI” is a broad term. In practice, such systems mainly accelerate research and help preliminarily test ideas.

Scientists believe that approved “AI drugs” for humans are still a long way off. Experts highlighted the complexity of obtaining regulatory approvals and the lack of publicly available high-quality experimental data.

Frank von Delft, a professor at Oxford University, stated that potential drugs still need to be tested for efficacy, toxicity, and suitability for storage and delivery, despite AI assistance. Clinical trials are the most expensive, he emphasized.

In July, the AIRI research institute introduced the GENATATOR tool, based on a neural network, for gene mapping by DNA sequence.