Artificial intelligence firm Anthropic reported on Thursday that it successfully disrupted multiple state-sponsored surveillance operations that were utilizing its AI models. According to the company, these campaigns were active between January and July of this year and were designed to monitor and target specific groups, including diaspora communities and pro-democracy activists from Hong Kong.

Anthropic’s findings highlight an emerging trend in which national governments are increasingly leveraging generative AI tools to enhance their surveillance capabilities. The company stated that the identified operations originated from China, Iran, and various locations in West Africa. The primary objective of these campaigns, according to Anthropic, was to track dissidents and minority populations, raising significant concerns about the potential for AI to be weaponized against vulnerable groups.

While the report identifies the geographic origins of these operations, it does not provide specific details on the exact methods used to bypass safety guardrails or the specific identities of the state actors involved. The company emphasized that it has implemented measures to prevent its models from being used for such purposes in the future, though the incident underscores the ongoing challenge of preventing the misuse of powerful AI tools by sophisticated state entities. This report serves as one of the first major public disclosures by an AI developer regarding the active abuse of their systems for state-sponsored espionage and social control.