The artificial intelligence company Anthropic has identified several countries that attempted to use its Claude AI tool for malicious purposes, including propaganda, drone development, and potential biological weapons research, according to a report obtained by The New York Times. The company says it blocked the product in all detected attempts.

The report lists Russia, China, Iran, and Yemen among the nations cited. It details a range of malicious uses, including suspected monitoring of exiled dissidents by China and Iran.

A prominent case involved Russia, which allegedly used Claude to fabricate propaganda disguised as journalistic reports aimed at influencing elections in neighboring Moldova. The AI was also reportedly used in attempts to develop military drones.

Anthropic’s disclosure underscores the growing challenge of preventing the weaponization of advanced AI by state actors.