Artificial intelligence models could be misused to facilitate biological threats, Anthropic warned in its latest report.
The company detailed potential misuses of its AI models in a sprawling 154-page report covering activity from December 2025 through August, including cases involving biological misuse, surveillance, scams and fraud.
"We’re publishing this work because we believe we have a responsibility to disclose malicious misuse of our services," the company said. "As models become increasingly capable, their risks will increase, unless AI developers and society’s defenders act to make them safer."
Anthropic described several examples in which bad actors tried to go around its controls and said the company said it then subsequently banned users' accounts. Anthropic said it identified cases where, in one case, a researcher outside the U.S. used Claude in their research on bird flu.
"All the evidence we have points to this being a research plan in its very early phases. However, the details of the plan show that this was research into a pathogen with enhanced pandemic potential," the company said.
The company noted that the research raised a difficult question of dual use: The same work could potentially be used to better understand a virus or to modify it in ways that could increase its threat.
Anthropic also said it found a user used Claude to research grant applications and ways to genetically change an organism or create new "biological property" on a mosquito-borne virus. That user caused concern because the institution affiliated with the grant application "was also a cause of concern" because it was set to be used at a military research institute.
"We hope that sharing these early insights with the public helps inform governments, the industry, and the general public on the nature of these risks, and the safeguards that are necessary for ensuring the safe deployment of AI models," the company said.