Anthropic has announced enforcement actions against the use of artificial intelligence (AI) technology as a weapon in intelligence operations by governments worldwide. Between January and July, the company identified cases where authorities in China, Iran, and West African nations increasingly utilized AI to surveil and monitor ethnic minorities and government opponents.
In these surveillance operations, Hong Kong pro-democracy activists, Tibetan and Uyghur members were primary targets, as were Iranian minorities. Iran-backed groups developed methods to identify and discriminate against people through social networks, while Mali's security authorities used Claude to develop their intelligence software. Beyond these efforts to replace engineers and attempts to create biological weapons and conduct online fraud, Anthropic successfully prevented these activities in advance.
Additionally, Chinese AI companies Moonshot and Deepseek allegedly stole information from Claude to improve their own models. Moonshot reportedly used over 5,380 fake accounts within ten days and sent nearly 300,000 user requests to Anthropic. The report strongly warns that these unauthorized acquisitions of technology from American companies without permission could also compromise users' personal data security.
Ref: Bangkokpost
