AI Surveillance Used by China and Iran Against Dissidents

• •Anthropic disrupted state-sponsored AI surveillance targeting dissidents.
• •Chinese firms allegedly misused Anthropic's AI for intelligence operations.
• •Sensitive user data was potentially exposed through fraudulent accounts.

Anthropic reported that it disrupted multiple state-sponsored surveillance operations using its AI models from January to July 2026. These operations targeted ethnic minorities and dissidents, particularly pro-democracy figures in Hong Kong, Tibetan and Falun Gong communities, and Iranian opposition abroad. Iranian actors developed methods to identify individuals via social media, while a contractor in Mali used AI to enhance intelligence gathering. Anthropic accused Chinese firms Moonshot and Deepseek of misusing its AI, relaying nearly 300,000 requests through fraudulent accounts. Sensitive user information was potentially exposed, violating privacy agreements. The report also noted attempts to design weapons and conduct questionable biological research. The intent behind some biological research was unclear, involving scientists working on potentially harmful pathogens.