OpenAI has again disrupted a Russia-origin influence operation, dubbed "Stop News", that used AI to generate content targeting Africa and the UK. This activity, consistent with an operation run by a marketing company, was first detailed in OpenAI's October 2024 report and continued into 2025.
The operation attempted to use OpenAI's models to create content for social media and websites posing as news outlets. OpenAI's investigation also identified efforts by this actor to use other companies' AI tools for video generation, which were then posted on YouTube and TikTok.
Key Points
- OpenAI banned accounts linked to the "Stop News" operation, which originated in Russia.
- The operation used OpenAI's models to generate influence content targeting Africa and the UK.
- Activity in 2025 focused more on text generation, with image generation being relatively rare, a shift from 2024.
- ChatGPT was used to generate scripts and descriptions for news-style short videos, including translation into French.
- Operators used OpenAI models to generate video prompts for other AI models, breaking Russian scripts into two-sentence snippets.
- Content criticized the roles of France and the United States in Africa while praising Russia, and also criticized Ukraine.
- The operation generated promotional materials for commercial companies in Russia, including ads for unlicensed online gambling.
Context
According to OpenAI, the "Stop News" operation's activity in 2025 appeared to be an attempt to continue an earlier operation following disruptions by OpenAI and industry peers. While the 2024 "Stop News" activity was noted for its prolific use of imagery, the 2025 activity showed a significantly lower proportion of image generation, focusing more on text. This aligns with an assessment by the French digital agency, VIGINUM, that the use of AI-generated images has drastically fallen since reports from Meta and OpenAI.
Why It Matters
This disruption highlights the ongoing challenge of malicious actors adapting their methods to continue influence operations despite platform interventions. The shift from image-heavy content to text and video script generation, and the use of multiple AI tools, indicates evolving tactics in the misuse of AI for influence campaigns.
What To Do
- Note the shift in influence operations from image-heavy content to text and video script generation.
- Watch for the use of AI models to generate video scripts and prompts for other AI tools.
- Compare the engagement levels of influence content, as this operation's content gained limited views despite complex AI use.
- Review OpenAI's October 2024 and October 2025 reports for detailed case studies on influence operations.