Already a subscriber? Make sure to log into your account before viewing this content. You can access your account by hitting the “login” button on the top right corner. Still unable to see the content after signing in? Make sure your card on file is up-to-date.
OpenAI has announced that it has banned multiple ChatGPT accounts it believes are linked to foreign governments attempting to misuse its AI tools for surveillance, covert influence operations, and other nefarious shit.
Getting into it: According to the report, multiple accounts believed to have been linked to the Chinese government used ChatGPT to generate proposals for large-scale surveillance initiatives. One request involved drafting a plan for a system that could analyze the travel movements and police records of Uyghur Muslims and other so-called “high-risk” individuals. In another instance, a user asked ChatGPT to help create promotional materials for a tool designed to monitor political and religious speech across social media platforms like X and Facebook. OpenAI stated that while these activities appeared to be conducted by individuals rather than formal institutions, they offered a rare glimpse into how authoritarian actors are seeking to co-opt AI tools for repression.
OpenAI also found that several Chinese-language accounts used ChatGPT to explore phishing and malware operations. These users prompted the model to help automate phishing workflows and sought ways to integrate China’s DeepSeek AI model into their activities.
On the Russian front, OpenAI identified accounts involved in a covert influence operation that used ChatGPT to draft social media scripts, generate SEO-optimized text, translate content, and create prompts for AI video tools. These materials were then used to spread politically motivated narratives on various social media platforms. The content promoted pro-Russia messaging, criticized the US and France for their involvement in Africa, and attacked Ukraine, the EU, and NATO.
The videos appeared to be shared on both Instagram and X.
