OpenAI announced on Thursday, October 8, 2026, that it has banned two separate influence operations based in Russia and Iran that used ChatGPT to generate propaganda and operate deceptive organizations. The company reported that these operations used artificial intelligence to create fake news stories, operate a fraudulent think tank, and manage a network of false journalist personas to promote the political agendas of both nations.
The disruptions follow previous reports from OpenAI in 2025 that foreign adversaries were using AI models to support cyber operations. The current findings come amid recent data from a Reuters/Ipsos survey indicating that 57 percent of respondents believe the Trump administration is not sufficiently addressing risks posed by AI, while 54 percent expressed similar concerns regarding Congress. The administration has generally supported voluntary safeguards, recently obtaining a "morally-binding" self-policing pledge from firms including OpenAI, Nvidia, and Meta.
OpenAI identified the Russian operation as "Dark Clark," which primarily targeted Latin America to undermine the reputation of Ukraine and influence local politics in Argentina and Bolivia. The Iranian operation, dubbed "Bogus Bylines," used seven false journalist personas to pitch approximately 100 articles to small and medium-sized online outlets worldwide. These stories frequently focused on the ongoing U.S. war with Iran, which began earlier in 2026.
OpenAI used the Breakout Scale—a 1-to-6 metric for measuring the impact of influence operations—to rate the Russian campaign at Category 5, the highest reach the company has recorded since it began tracking such activity. The Iranian campaign reached Category 4. For the average reader, this means AI-generated content is no longer confined to social media bot accounts but is successfully penetrating mainstream media outlets. A resident in Peru, for example, might have encountered news reports about local schools holding events for Ukrainian nationalists that were later revealed by OpenAI to be based on fake emails and planted stories designed to create diplomatic tension between Ukraine and Poland.
OpenAI stated that these disclosures are intended to assist regulators and industry peers in identifying vulnerabilities in AI systems. The company noted that the Russian operators used virtual private networks (VPNs) to bypass bans on accessing the models from within Russia. While OpenAI has now suspended the accounts and shared information with relevant authorities, the company did not provide a specific deadline for new security implementations. The Trump administration's current policy continues to rely on voluntary reviews of AI systems by the tech companies themselves rather than federal legislation.