AI-Powered Espionage Activities Blocked
AI company Anthropic announced that it halted state-backed espionage operations originating from China, Iran, and West Africa between January and July.
Anthropic announced that it has blocked state-backed espionage operations in which AI models were used to surveil dissidents and ethnic minorities.
Operations Between January and July
In a report published by Anthropic, it was reported that multiple state-backed surveillance operations detected between January and July were terminated.
Targeted Communities and Regions
It was stated that these surveillance activities originated from China, Iran, and West Africa, and historically targeted diaspora and dissident communities.
In this context, it was noted that pro-democracy figures in Hong Kong, as well as Tibetan and Falun Gong communities and Iranian minorities, were targeted.
Espionage Methods and AI Usage
It was stated that Iranian actors developed a method to identify individuals through social media accounts, while a contractor working for Mali's national security officials used Claude to design intelligence software.
Other Abuse Cases
Anthropic also reported that it blocked attempts at weapon design, suspicious biological research, and romance scams, while accusing Chinese developers such as Moonshot and DeepSeek of deceptively using Claude to improve their own models.