Anthropic Faces Backlash After Claiming Claude Made a Biological Discovery
Biologists and drug executives are pushing back against Anthropic’s claims that an ensemble of Claude agents made an original scientific discovery, warning that AI labs are overhyping routine computational filtering.
Anthropic Claims Its AI Lab Made a Biology Breakthrough
Anthropic announced that a molecular biology lab it established earlier this year has yielded its first scientific discovery. In the facility, instances of Anthropic’s Claude models read biological literature, formulate conjectures about complex biological challenges, and hand off findings to human scientists who carry out laboratory experiments.
Anthropic framed the achievement as an AI-driven discovery, comparing the findings to the initial mechanisms that led to the development of CRISPR gene editing. However, the claim that the AI system itself made a scientific breakthrough has triggered immediate pushback from life-science researchers, who argue that the company is confusing computational assistance with genuine discovery.
What Claude Actually Found—and Why Scientists Are Skeptical
Anthropic’s announcement centered on an automated search through vast libraries of sequenced DNA. A network of 950 Claude agents operated for 21 hours, screening data until flagging an uncatalogued, repeating pattern surrounding a known enzyme.
Biologists quickly took issue with Anthropic's characterization of the result. Lucas Harrington, a biologist whose critique went viral and received public backing from the chair and CEO of pharmaceutical giant Eli Lilly, pointed out that identifying gene repeats is standard legwork rather than an end result. Harrington argued that locating an unusual cluster of repeats and genes is typically the easy phase, while determining the biological mechanism and function of the system constitutes the actual breakthrough.
The novelty of the finding has also been challenged. Mario Rodríguez Mestre, a biologist at the University of Copenhagen, reported that his research group had already uncovered the exact same pattern. Because Mestre had routinely consulted Claude during his research, he questioned whether Anthropic’s team had extracted insights from his user interactions. Anthropic denied utilizing his prompts to inform the project, but Mestre stated that he has discontinued all use of Claude.
While filtering hundreds of thousands of candidate sequences down to viable leads represents substantial scientific labor, critics stress that such computational sorting functions like a high-powered instrument rather than an independent researcher.
The Escalating PR Race Between Anthropic and OpenAI
The controversy reflects a broader trend among leading AI labs repositioning their software from scientific tools—comparable to microscopes or supercomputers—into autonomous discoverers. This positioning comes as Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman compete aggressively to prove their models can independently advance frontier science.
OpenAI sparked a similar debate earlier in the month after asserting that an ensemble of its own agents solved a million-dollar mathematics problem. That announcement was soon followed by pushback from mathematicians who argued the specific problem solved was not central to the field, alongside accusations that the models drew on prior, uncredited work from a human mathematician.
Harrington cautioned that AI developers need to establish higher standards for what qualifies as a discovery. Without a strict threshold, scientists warn, the race to claim PR milestones will only fuel skepticism and obscure the moments when artificial intelligence actually uncovers fundamentally new natural mechanisms.
