Anthropic has set up its own laboratory in the San Francisco Bay Area to conduct physical biological experiments and is developing ways to use Claude to automate laboratory procedures. Two sources familiar with the matter told Reuters about this, and the existence of the laboratory was confirmed by Eric Kauderer-Abrams, head of the life sciences division. The company has no plans to conduct clinical trials of its own drugs.
Briefly about the main points
- Anthropic has confirmed that it operates its own biological laboratory.
- The company is moving from modelling to physical experiments.
- Claude could be involved in supervising laboratory work.
- Anthropic’s focus is on pre-clinical tools, rather than clinical trials.
- The company combines technological development with the management of biosafety risks.
From modelling to testing in a real-world laboratory
Kauderer-Abrams He stated that Anthropic’s biological work is no longer limited to in silico research — computer modelling and data analysis. «The ultimate test for biology remains, and for some time to come will remain, real laboratory work,» he explained.
The company carries out some of the experiments itself, whilst others are conducted in collaboration with external partners. However, the laboratory was not set up specifically as a facility for drug development.
AI should control the robots, but under human supervision
One area of application is the use of Claude to control robotic systems that will carry out scientific experiments with minimal human intervention. According to a Reuters source, Anthropic is currently working on precisely this model for organising laboratory work.
The company has confirmed that this technology is at an early stage of development. Anthropic believes that automation has the potential to speed up a wide range of scientific procedures, but emphasises that human oversight is essential for safety reasons.
Focus on complex diseases and multi-target antibodies
Anthropic states that it wishes to assist in the research into diseases that the traditional pharmaceutical industry may consider to be insufficiently profitable. According to the company, AI It may speed up the search for potential treatments for complex diseases.
Among the areas mentioned are complex molecules — bispecific and trispecific antibodies — capable of interacting with several targets simultaneously. The company acknowledges that even after a promising molecule has been identified, it can take years before the finished drug reaches the market.
The company limits its role to pre-clinical research
Anthropic has no plans to move on to clinical trials at this stage. Kauderer-Abrams explained that the company does not wish to compete directly with pharmaceutical and biotechnology manufacturers that bring medicines to market.
Its stated focus is on pre-clinical research and technologies capable of accelerating the work of other players in the pharmaceutical industry. In other words, Anthropic positions itself as a developer of tools and scientific infrastructure, rather than a manufacturer of medicines.
Partnerships in the pharmaceutical sector and the assessment of biosafety risks
Anthropic is also expanding its work in the life sciences beyond the laboratory: the company has launched the Claude Science software, acquired the start-up Coefficient Labs and appointed the CEO of Novartis to its board of directors Vasa Narasimhan. It provides AI tools and services to Genentech, Bristol Myers Squibb and Novo Nordisk.
Anthropic has previously reported instances where its systems could have been used to aid in the development of biological weapons, which has sparked expert debate about the extent of such risks. The company states that the development of models for the life sciences requires equal attention to both benefits and safety.
Anthropic has not disclosed the detailed results of its own biological research, nor has it announced the development of any specific drug. It is currently developing tools designed to combine AI, robotics and biological research.
