Anthropic Says Claude Autonomously Discovered a Crispr-Like Enzyme System

Anthropic says its AI Claude has "autonomously discovered" a new enzyme system similar to the machinery behind the powerful gene-editing tool Crispr. The announcement marks the first result from Anthropic's newly launched wet lab, and it lands as the company prepares to go public.

The claim carries weight beyond a single experiment. Crispr's discovery reshaped genetic engineering and earned a Nobel Prize; if AI systems can surface comparable biological machinery from raw genetic data, the stakes for both research and the commercial case for AI in science are substantial. The result also serves as an early test of Claude's usefulness for scientific discovery, a metric investors and customers will scrutinize ahead of an IPO.

According to Anthropic, Claude found the enzyme system by searching through a massive database of DNA sequences. The company says its scientists' involvement was limited to the initial prompt and the lab work that followed.

The scale of the search was considerable. Over 21 hours, Anthropic said nearly 950 Claude agents worked through 210 million tokens before one agent spotted an unusual repeating pattern in the data. That repeating structure pointed to the novel enzyme system, echoing the repetitive DNA sequences that first tipped researchers off to Crispr's bacterial defense machinery.

The distinction Anthropic draws — that the discovery was autonomous rather than human-guided — matters for how such results are evaluated. Anthropic frames the outcome as evidence that large language models can do more than summarize or draft text: they can drive an open-ended scientific search across vast datasets with minimal human intervention.

The wet lab itself is new. Anthropic launched the facility to pair Claude's computational searches with physical experiments, closing the loop between hypothesis and validation. This enzyme system is its first published output, which makes it a proof of concept for the lab's broader ambition: positioning Anthropic in the emerging market for AI-driven scientific discovery, where competitors including Google DeepMind have staked claims with tools like AlphaFold.

The company's timing is deliberate. Anthropic is preparing to go public, and demonstrations that Claude can deliver concrete scientific results strengthen the narrative that its models create value beyond chatbot applications. An autonomous discovery likened to the origins of Crispr is exactly the kind of result that supports premium positioning in enterprise and research markets.

Questions remain that the announcement does not answer. Independent verification of the discovery, peer review, and details of the enzyme system's function will determine whether the comparison to Crispr holds up. Crispr itself took years of follow-up research before its full significance became clear.

For now, Anthropic has shown that nearly 950 agents, 21 hours, and 210 million tokens can surface a pattern humans had not flagged. Whether this becomes a repeatable path to genuine scientific breakthroughs — or an isolated headline — will depend on what the wet lab produces next.