Anthropic has confirmed that it operates a wet biology lab in the Bay Area where it integrates its advanced AI models to conduct physical biological experiments. This move follows its acquisition of the stealth biotech startup Coefficient Bio in April 2026 and signals the company’s commitment to applying AI beyond virtual simulations by validating biological theories through real lab work. Unlike drug discovery efforts, Anthropic emphasizes that the lab's main focus is fundamental biology, and it collaborates with external partners while aiming to avoid direct competition with its pharmaceutical customers, such as Novo Nordisk.

Recently, Anthropic announced a significant discovery made in this lab: a novel enzyme system with functions akin to CRISPR, the famed gene-editing technology used to modify DNA. This previously unknown enzyme complex, found within the DNA of bacteriophages (viruses that infect bacteria), reportedly can perform molecular operations like cutting, copying, and pasting DNA segments. While this discovery requires further validation from the scientific community, CEO Dario Amodei highlighted that much of the identification work was led by Anthropic's AI model, Claude, within just 21 hours using thousands of data agents and processing over 210 million tokens.

The announcement comes amid a period of heightened concern within the AI industry regarding the risks of unchecked AI development. Some of Anthropic’s own researchers have publicly voiced worries about existential risks posed by AI, including the potential for bioterrorism, a threat Amodei has identified as a priority risk. Despite these fears, Anthropic continues its biological research, insisting that all lab experiments are conducted by human scientists under biosafety level 1 and 2 standards and that the AI system Claude has not been given autonomous control over lab equipment.

Anthropic’s engagement in AI-driven biology research is part of a broader trend where institutions like Stanford and UC San Francisco also leverage AI to advance enzyme design and gene-editing techniques. While other companies, such as Google with AlphaFold, have pioneered AI tools for biology, Anthropic’s integration of a dedicated wet lab marks a notable step. Amodei envisions a future where AI might safely perform such experiments autonomously but maintains that human oversight remains critical at present, reflecting the company’s cautious approach in balancing innovation with ethical responsibility.