OpenAI reported that it banned a ChatGPT account tied to a coordinated influence operation it named “No Bell.” The company said the account produced long‑form articles and social media content about geopolitics in sub‑Saharan Africa, and that the investigation began after a lead from Meta.
Operation and tactics
According to OpenAI, the account primarily generated social posts and commentary articles in English, sometimes using Russian‑language prompts that the user attributed to a manager. The user requested outputs without em‑dashes and in the style of a human journalist to reduce the appearance of AI generation. OpenAI also observed the operator experimenting with multiple models and image tools, at times asking for Gemini image prompts and edits.
Many articles carried the byline “Dr Manuel Godsin,” presented as a PhD from the University of Bergen and an employee of the “International Centre for Political and Strategic Studies.” OpenAI reported it could not find evidence that such a person exists: searches of the Norwegian National Research Information Repository (NVA) and the Bergen University Library produced no matches. Online traces of the author photo matched a profile of a St Petersburg law student, OpenAI said.
Content and targets
OpenAI said the operation typically praised Russia while criticizing Ukraine, the United Kingdom and the United States. Some pieces were personal in tone and focused on Volodymyr Zelenskiy or Donald Trump. Other articles alleged that Rheinmetall used a South African subsidiary to evade arms export controls and accused the British NGO Crisis Action of fomenting protests in South Africa, per the report.
Content generated by the account was published on a range of African news websites and posted to Facebook Pages posing as local outlets in countries including South Africa, Ghana, Kenya and Angola. Meta removed several of those Pages; transparency data for some showed admin locations in Russia and Ukraine. One Page had about 3,000 followers before takedown, while others had minimal reach.
OpenAI assessed the operation toward the low end of Category 4 on its IO impact Breakout Scale, reflecting some placement in mainstream African media but limited social media influence. The case appears in OpenAI’s February 2026 report on disrupting malicious uses of AI, and OpenAI confirmed it banned the involved ChatGPT account.
Original source: OpenAI News