Anthropic's 950 agents found a pattern, not a breakthrough

Anthropic announced last Wednesday that it had launched a molecular biology lab earlier this year, where Claude agents read and conjecture about hard biology problems while human scientists run experiments on what they report. The company said the system of 950 agents made its first discovery after 21 hours, flagging a repeating pattern around a known enzyme. In its announcement, Anthropic described the pattern as reminiscent of what led to the gene-editing technology CRISPR.

That framing drew pushback. Lucas Harrington, a biologist, wrote a viral post arguing that finding a weird cluster of genes and repeats is often the easy part, and that the hard part is figuring out what the system actually does. His post was subsequently endorsed by the chair and CEO of Eli Lilly. Separately, Mario Rodríguez Mestre at the University of Copenhagen said his team had already discovered this particular pattern, and he is stopping all use of Claude, though Anthropic denies his team's work informed the finding.

The episode points to a broader tension: AI companies are presenting their systems as making discoveries themselves, rather than as tools like microscopes or supercomputers. That framing may make people more skeptical of genuine progress when it happens, as seen when OpenAI said its agents cracked a million-dollar problem in mathematics and skeptics questioned whether the result was the one mathematicians care most about. Harrington suggested AI companies should set the bar high now, so that when an AI actually discovers a fundamentally new biological mechanism, everyone appreciates how big a deal it is. Whether that standard takes hold may depend on how the dispute over credit for this pattern is resolved.