OpenAI has disrupted a revival of the Russian influence operation “Stop News,” a campaign it first exposed and banned a year ago. This time, the operators weren’t just churning out AI-generated articles. They used ChatGPT to script propaganda videos, generate SEO-friendly descriptions, and even produce prompts for other AI video models to create realistic-looking newsreaders.
Much like the original operation, the content targeted audiences in Africa and the UK through a network of sites posing as legitimate news outlets. The narratives followed a predictable Kremlin playbook: French-language pieces slammed France and the U.S. for their roles in Africa while burnishing Russia’s image, while English-language content attacked Ukraine and its allies. That part of the script hasn’t changed.
What’s different is the production value. The operators fed lengthy Russian text into ChatGPT, commanded it to spit out a video script, then broke that script into bite-sized snippets before ordering the model to generate video prompts for each one. The result? A series of TikTok and YouTube videos featuring an AI-generated newsreader whose appearance evolved several times throughout 2025. VIGINUM, the French digital agency, previously noted that the use of AI imagery dropped sharply after the initial industry crackdowns. OpenAI’s latest data confirms that shift—image generation was significantly down, while text and script generation surged.
Some of the promotional content bordered on reputation laundering. The operators asked ChatGPT to write Google ad copy and articles that grouped their fake “Newstop Africa” brand alongside real news outlets, a cheap trick to borrow credibility. A small but telling detail: one French caption in a video contained a grammatical error that wasn’t in ChatGPT’s original translation, suggesting it was transcribed from audio and never proofread by an actual French speaker. These aren’t master propagandists. They’re just persistent.
