Anthropic Says Governments Are Using Claude for Surveillance

Anthropic says governments and state-linked organisations are increasingly using its Claude AI models to support surveillance activities. In its September 2026 threat report, the company described cases involving actors connected to China, Iran and Mali, as well as commercial surveillance firms, and said the activities violated its policies.

According to Anthropic, some operators used Claude to analyse information, prepare intelligence reports and support surveillance systems. In one case involving Mali, the company said Claude was used as part of the development of a large-scale communications surveillance platform. Anthropic said it disrupted the activity and banned the associated account.

The report also describes cases involving Chinese and Iranian state-linked actors. Anthropic said some operations focused on monitoring dissidents, activists and diaspora communities, while other efforts involved building or improving surveillance infrastructure. The company said it banned the accounts involved and strengthened its systems for detecting similar misuse.

Anthropic’s findings highlight a broader concern about the growing ability of AI to reduce the amount of human expertise needed for complex technical and analytical work. The company said AI is increasingly being incorporated into government security operations, meaning capabilities that previously required larger teams of specialists can potentially be carried out more efficiently.

The revelations have intensified debate over how AI companies should prevent their technologies from being misused by governments, private organisations and other actors. Anthropic says it will continue improving safeguards and sharing relevant information with authorities and industry partners. The report also demonstrates that controlling AI misuse is becoming an international challenge involving privacy, civil liberties and the responsible deployment of increasingly capable systems.