Anthropic reports AI misuse attempts for weapons, pathogens, and surveillance

A threat‑intelligence report released Thursday by Anthropic, the artificial‑intelligence firm behind several high‑profile language models, details how its technology has been exploited by a range of malicious actors. The 154‑page document cites attempts by criminals, state‑sponsored groups, spyware vendors, scientists and propagandists to use Anthropic’s models for designing missiles and bombs, engineering lethal pathogens and surveilling political dissidents. Anthropic emphasized that the cases highlighted represent “the most notable and novel threat activity” it has identified, and said the disclosure reflects its responsibility to reveal malicious misuse of its services.

The report follows a recent internal controversy in which a former Anthropic employee resigned, warning that the company’s models could contribute to human extinction by 2030. While the employee’s claim remains unverified, Anthropic’s findings illustrate concrete instances of misuse that extend beyond speculative risk. The company has not indicated any immediate changes to its product offerings, but the publication underscores growing concerns within the tech industry about the dual‑use nature of advanced AI and the need for robust safeguards against its exploitation.