Anthropic disrupts state-sponsored surveillance using Claude AI in China, Iran, Yemen

Anthropic released a threat intelligence report detailing multiple cases of misuse of its Claude models between January and July, disrupting state-sponsored surveillance operations originating in China, Iran, and West Africa targeting ethnic minorities and dissidents. The report also revealed attempts by Yemeni threat actors to use Claude for guided weapons development, Chinese AI firms extracting tokens from U.S. models, and efforts to develop bioweapons. Nextgov/FCW reported that U.S. intelligence agencies warned of Chinese companies including DeepSeek and Moonshot AI conducting large-scale model distillation. The incidents prompted broader calls from U.S. lawmakers for new AI governance rules, with Anthropic CEO Dario Amodei calling for a coordinated slowdown in frontier AI development.