OpenAI bans ChatGPT accounts tied to surveillance and malware

OpenAI said on Tuesday it has suspended and banned a number of ChatGPT accounts after investigators found they were attempting to use the service to plan social media surveillance, phishing campaigns, malware development and other activities that violate the company’s policies. The actions are described in OpenAI’s latest public threat report, which outlines investigations into accounts the company said were linked to actors operating in China, North Korea, Russia and several Southeast Asian and African countries.

According to the report, several Chinese-language accounts used virtual private networks to connect to ChatGPT from within China and sought assistance drafting proposals and promotional material for a social media listening tool designed to monitor platforms including X, Facebook, Instagram, Reddit, TikTok and YouTube for content defined by operators as extremist or related to ethnicity, religion or politics. OpenAI said it also banned an account that drafted a proposal for what it described as a High-Risk Uyghur-Related Inflow Warning Model intended to track Uyghur movements and cross-reference them with police records. The company said the banned accounts did not...