OpenAI released a report detailing the banning of two clusters of ChatGPT accounts likely originating from China, which were used to generate fake social media comments and images to manipulate public debates on US AI data center energy costs and tech tariffs. This incident highlights generative AI as a new tool for geopolitical information warfare and underscores the platform's responsibility in identifying and disrupting automated influence operations.
Influence Operations
影响力操作
Systematic efforts to affect public opinion through media or online channels
Curated Intelligence
OpenAI disclosed the 'No Bell' malicious use case in its latest report, banning a suspected Russian-origin ChatGPT account. The account generated long-form articles and social media posts on African geopolitics to influence public opinion via fake news outlets and antagonize US political teams. This case highlights the risks of LLMs in low-cost, large-scale information manipulation and platform detection mechanisms.
OpenAI banned accounts linked to the Rybar network, suspected of originating in Russia. These accounts used AI to conduct multilingual influence operations across websites and social platforms.
OpenAI disclosed a 'Tort Report' abuse case in its threat intelligence report, where Vietnamese accounts used models to generate comments for filing malicious reports against independent media on Facebook and YouTube. Although the activity resulted in no takedowns (rated as Category 1 impact), it highlights an emerging risk of LLMs being weaponized to automate false reporting and suppress dissent.
Meta banned an influence network operated by Israeli firm STOIC that used Meta's own AI models to generate articles and social media comments. Despite targeting multiple countries, the operation attracted negligible engagement, leading to its nickname 'Zero Zeno'.