Peruvian and Ecuadorian schools booby-trapped, fake audios, a façade research center: OpenAI details a campaign linked to the Politology network, and an Iranian operation of fake journalists.
OpenAI announced on October 8 that it had banned two clandestine influence operations that relied on ChatGPT to support front entities. The first, of Russian origin and nicknamed "Dark Clark", is the first to reach category 5 on the Breakout scale (from 1 to 6) among the thirty or so operations exposed by the company in two and a half years. Its operators, connected by VPN and writing mainly in Russian, wrote internal reports with ChatGPT on three projects: denigrating Ukraine and curbing its military recruitment, influencing Bolivian and Argentine political life, and managing a "Social Research Center" run without their knowledge by local employees under the false identity "Mia Clark".
Schools targeted
According to these reports, the team spoofed the address of the Lima regional directorate of education in May to push Peruvian schools to organize events around Stepan Bandera, and then in June pressured Ecuadorian schools to pledge allegiance to President Noboa and Erik Prince; OpenAI found corresponding articles and the reaction of a Polish MEP. The group also released fake recordings, including one announcing a water cut in La Paz. Open-source researchers link these actions to the "Politology" network, presented as a successor to Wagner — although OpenAI notes that the operators also took credit for incidents in which they had not taken part.
The second operation, Iranian and classified as a category 4, had seven fake Western journalists sign nearly 100 articles placed in a dozen online media outlets, from July 2025 to October 2026. OpenAI says it has passed the information on to the relevant authorities.
Read the full article on the original source
Go to the source →