Governments Are Using Claude to Automate State Surveillance
Anthropic has revealed that state-run surveillance operations in China, Iran, and Mali are actively using its Claude AI models to streamline civilian monitoring and automate target selection.

Key takeaways · 3
- 01
Iranian actors used Claude to develop a malicious Firefox extension for social media identity harvesting.
- 02
A Chinese intelligence office replaced multiple analyst teams with a single AI-assisted office.
- 03
Anthropic banned a pre-operational account attempting to build a commercial surveillance platform for the Persian Gulf.
Global Surveillance Operations
Anthropic released a threat report on Thursday stating that state-run surveillance operations are using Claude to streamline their activities. [1] The AI lab found people linked to governments in Mali, China, and Iran utilizing various Claude models for these purposes. [1] In Mali, a consultant for national security authorities used Claude to create a system that collects data from mobile operators and builds personal dossiers. [1] Furthermore, Anthropic disclosed that Iranian actors used the AI to build a malicious Firefox extension that harvested user identities from social networks. [1]
Automated Target Selection
According to the report, governments have used Claude to analyze information and select targets for surveillance. [1] In China, a religious affairs intelligence office reduced its operations from numerous analyst teams to a single office that produces thousands of investigations per month using an AI assistant. [1] The Chinese government also deployed Claude to assess the political sensitivity of social media data and select targets for interventions like coercive questioning. [1] Additionally, Anthropic noted that one Chinese state security bureau created an internal manual for using AI in surveillance operations. [1]
What it means
The integration of large language models into state surveillance apparatuses demonstrates that generative AI is moving from an experimental technology into everyday bureaucratic infrastructure. The reported transition of Chinese intelligence operations from multiple teams to a single automated office illustrates a massive reduction in the labor required to monitor civilians. While Anthropic successfully detected and banned a commercial surveillance platform targeting the Persian Gulf before it launched, the widespread use of these models by state actors highlights severe security challenges. What the sources don't address: Whether Anthropic plans to implement stricter geographic or systemic access controls to prevent state actors from bypassing their terms of service.
The adoption of frontier AI models by state security apparatuses lowers the technical and labor barriers to mass surveillance. As these models transition into everyday bureaucratic tools, AI companies face increasing challenges in policing malicious use across borders.
Why it matters
Put this to work — one session a day, built for your industry.
Create a free account for a daily session — eight questions and one real-work challenge, on the news that affects your role.
Start freeHow this developed
11 September 2026
Governments Are Using Claude to Automate State Surveillance
11 September 2026
Event created from source cluster.