209/68 Wednesday, June 11, 2025
OpenAI has announced that it has suspended numerous ChatGPT accounts found to be linked to state-sponsored threat actors, particularly from Russia and China. The move follows internal investigations revealing that these accounts were being used to support malicious activities worldwide—including employment scams, social engineering, and cyber espionage. According to OpenAI’s report, these accounts were involved in generating malware code, conducting social media automation, and researching U.S. satellite communication technologies, among other sensitive topics.
Over the past three months, OpenAI’s investigative team has identified and disrupted several forms of abuse, including job scam schemes, impersonation to gain system access, and influence operations. Approximately 40% of these cases originated from China, with additional accounts linked to actors in Cambodia, the Philippines, Iran, and Russia. Many China-based accounts were used to generate massive volumes of social media content on sensitive topics such as the termination of USAID operations and U.S. domestic political conflicts. Cambodian-linked accounts used ChatGPT to create recruitment-style messages in multiple languages, while North Korean users reportedly used the tool to research technical evasion tactics to maintain covert remote access. Russian hackers were seen using ChatGPT to develop Windows-based malware, debug malicious code, and build command-and-control infrastructure.
OpenAI stated that some of the malicious activity was linked to APT5 and APT15, known Chinese state-backed threat groups. This crackdown aligns with OpenAI’s core mission to ensure that AI benefits humanity under reasonable safeguards, and to protect people from harm. “This includes preventing authoritarian regimes from leveraging AI to consolidate power, suppress citizens, or threaten other nations,” the company said, highlighting the need to prevent misuse in disinformation operations, child exploitation, fraud, spam, and malicious cyber activities.
