Anthropic's biology lab says it has already made a significant discovery, though it isn't saying what yet.
The company disclosed the finding while describing how it uses Claude in scientific research. Notably, Claude isn't running experiments unsupervised in that lab. Humans review the AI's suggestions before anything gets tested, and Anthropic didn't detail what the discovery actually is or when it might be published.
The disclosure doubles as a positioning move. Anthropic gets credit for a scientific result without inviting the harder questions about an AI model working unsupervised near pathogens or lab equipment. Any company building AI into biology research, whether for drug discovery or automated lab work, eventually has to answer how much autonomy it will give a model that can design its own experiments.
A big discovery with no data attached is, for now, mostly a claim.