Biologist alleges Claude AI used unpublished research in enzyme discovery
A Copenhagen University biologist alleges that Anthropic's Claude AI model may have used his unpublished enzyme research to make a discovery the company recently announced, raising questions about data use in AI training and intellectual property in scientific research.
What happened
Biologist Mario Rodriguez Mestre from Copenhagen University alleges that Anthropic's Claude AI model may have used unpublished research from his four-year study of enzyme systems (ARTs) to make a recent enzyme discovery that Anthropic announced.
Mestre states his team has regularly used Anthropic's AI models in their research for the past three years and provided important findings about the enzyme to Anthropic during that process. The core question, according to Mestre, is whether Claude reached its conclusion through independent reasoning or was partially directed by his team's earlier work.
Mestre expressed concern that information from his research could be used in training future versions of the model. Anthropic stated it is not aware of any prior published studies describing the ARTs system and that Claude was not trained on user input, with its molecular biology team lacking access to such records.
Reported impact
The allegation has raised broader questions about how much autonomous authority should be granted to AI systems in biotechnological research.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.