The latest

Anthropic has opened a wet lab in the San Francisco Bay Area, giving the Claude developer an in-house site for physical biology experiments as it deepens its life-sciences push. The company conducts some experiments internally and others with external partners, said Eric Kauderer-Abrams, its head of life sciences. A spokesperson said the facility is not specifically dedicated to drug discovery, while the wider effort partly targets treatments for rare and poorly served diseases.

Details

• Laboratory automation: Anthropic wants Claude to interact with robotic laboratory equipment and help execute experiments with limited human intervention. Kauderer-Abrams said laboratory work will remain biology’s ultimate test for a long time, describing AI-led automation as being at a very early stage despite its potential to accelerate scientific processes.
• Hardware framework: The company has introduced the Model Hardware Standard, a framework intended to let AI systems interact with physical equipment. The initiative supports Anthropic’s broader effort to connect its models with machines used in laboratories and other industrial settings.
• Preclinical focus: Anthropic plans to pursue preclinical research in disease areas that traditional pharmaceutical companies may find commercially unattractive because of market size or complexity. Kauderer-Abrams said the company is not currently pursuing clinical trials and intends to complement established pharmaceutical and biotechnology groups rather than compete in drug commercialization.
• Scientific investment: Life sciences is already among Anthropic’s largest investment areas by personnel and resources. The company acquired pharmaceutical-research tools startup Coefficient Bio, recruited specialists in protein and nucleic-acid characterization, and is seeking executives to expand procurement and laboratory operations.
• Pharma relationships: Anthropic is developing AI tools for Roche, Genentech, Bristol Myers Squibb and Novo Nordisk while building its own scientific capabilities. Some pharmaceutical customers are concerned the company could gain insight into competing research programs, despite protections for customer data. Anthropic says remaining outside clinical development for now partly addresses those concerns.
• Safety balance: Human oversight remains central to the laboratory strategy because of risks associated with applying increasingly capable AI systems to biology. Anthropic researchers have warned about catastrophic risks from advanced AI, while the company has said its systems could potentially be misused in biological-weapons development.

Background

AI developers and pharmaceutical researchers are increasingly using models to design molecules, predict biological interactions and automate experiments. Potential targets include bispecific and trispecific antibodies, which interact with multiple biological targets. Drug candidates still require laboratory testing, animal studies and human trials, with most failing during development. Alphabet’s Isomorphic Labs, spun out of Google DeepMind, is developing AI-based drug discovery approaches and moving candidates toward clinical development.

What’s next

The next indicator will be whether Anthropic expands its preclinical programs or changes its position on clinical development as laboratory capacity, specialist recruitment and partnerships grow.