Anthropic announced a molecular biology lab using 950 Claude agents. The agents flagged a repeating DNA pattern near a known enzyme. Anthropic called this a first discovery by its AI-powered lab. Biologists argued the pattern was not novel and the real work is understanding function. A Copenhagen team said it had already found the same pattern. Anthropic denied learning from that team. The biologist who found it stopped using Claude. AI firms present systems as making discoveries rather than aiding scientists. Critics say this framing obscures how science actually progresses. Skepticism grows over whether AI achievements meet scientific standards. OpenAI recently claimed its agents solved a million-dollar math problem. Critics then questioned if it was the right problem to solve. A mathematician accused OpenAI of using his work without credit.