Govt used Claude to track people: Anthropic

A government consultant in Mali utilized Anthropic's AI system, Claude, to devise a mass surveillance platform, according to the AI company's recent report. The consultant used Claude to create a system capable of intercepting mobile communications across Mali's mobile operators and compiling detailed information on targeted individuals. This case, detailed in Anthropic's September 2026 report, "Detecting and Countering Misuse of AI," was one of several instances of misuse of AI for surveillance purposes.

Anthropic identified and disrupted multiple surveillance operations between January and July, involving state-aligned actors, state-linked contractors, and commercial surveillance firms. The company explained that Claude was employed to engineer the software that enabled the intelligence gathering, rather than merely analyzing data that had already been collected.

According to Anthropic, this represented an example of AI being used in place of a larger engineering workforce, as a single consultant was able to use Claude to develop the surveillance system.

Beyond merely building technical tools, the surveillance operations extended to processing and organizing vast amounts of online information about individuals. For instance, an actor uploaded social media posts and employed Claude to convert them into structured records, which included details such as people's locations, demographic information, and political views. Additionally, an Iranian unit used Claude to analyze hundreds of thousands of social media posts, selecting 39 opposition accounts for monitoring.

Claude was also used in other surveillance operations, such as by a Chinese-aligned actor targeting Uyghurs in Syria. This actor, despite lacking Arabic language skills, utilized Claude to draft messages in a regional dialect, translate replies in real time, and role-play as an expert to assess the quality of the operation. Claude even formatted the results for suspected handoff to a case officer.

Furthermore, two Iranian units independently used Claude to address technical and usability issues within a central surveillance case-management system, with no shared code or personnel. Anthropic banned accounts associated with these operations due to their violation of its Usage Policy, which prohibits non-consensual surveillance and profiling, as well as the misuse of its services to infringe upon civil liberties and human rights.

The report highlighted Anthropic's concern that AI is increasingly being used not only to analyze information but also to construct the systems used for collection, process data on large numbers of individuals, and identify those who may be subject to further monitoring.