OpenAI Bans China-Linked Accounts for Seeking Uyghur Surveillance AI

By Alp Uyghur

WASHINGTON | October 8, 2025

OpenAI announced that it has banned multiple accounts linked to Chinese entities after they used ChatGPT to develop proposals and promotional material for large-scale social media “listening” probes and a system specifically aimed at AI surveillance. This action, detailed in the company’s latest threat report, confirms the illicit utilization of generative AI technology for state-affiliated espionage and tracking operations.

Open AI said: “Similarly, we banned a second user, likely connected to a government entity, who
asked ChatGPT to help write a proposal for what they described as a High-Risk
Uyghur-Related Inflow Warning Model (高危涉维吾尔关注人员流入预警模型). The
proposal described a tool that would analyze transport bookings and compare them
with police records in order to provide early warning of travel movements by people
classed as Uyghur-related and high-risk, an otherwise undefined category. The user
did not ask our model to help build such a tool, only to develop a general proposal; as
such, we are unable to independently confirm what technology they intended to use or
whether such a tool has been used by a Chinese government entity.”

Development of a “High-Risk Uyghur-Related Inflow Warning Model” designed to track the movements and communications of individuals associated with the ethnic minority group.The associated accounts were immediately banned, consistent with OpenAI’s policy against national security violations, including surveillance and the debugging of malicious code, OpenAI said.

Michael Sobolik, a senior fellow at the Hudson Institute, welcomed the move by OpenAI, saying, “encouraging to see OpenAI take the democratic and societal implications of AI seriously.”

Accounts connected to Chinese government entities and developers in Russia, Korea, Cambodia, Myanmar, and Nigeria.

This is not the first disruption of Chinese state-linked accounts by OpenAI, highlighting a sustained effort to leverage large language models for Uyghur tracking and other forms of political and ethnic monitoring.

Uyghur Times Staff

Leave a Reply

Your email address will not be published. Required fields are marked *

You May Like