OpenAI released a report exposing Russian and Iranian influence operations that used ChatGPT to plant fake content in legitimate news outlets rather than just social media. The Russian 'Dark Clark' operation became the first case rated Category 5 (highest risk) by OpenAI because it triggered official political responses in Latin America. Although AI was primarily used for internal reporting and language adaptation, fabricated audio and documents successfully provoked fact-checks and denials in Ecuador and Peru, marking a shift from online noise to tangible political interference.
Influence Operation
影响力行动
Activities systematically changing public perception or behavior via media or networks
Curated Intelligence
OpenAI has banned a suspected China-origin cluster accused of using AI to mass-generate social media content criticizing US data centers and AI infrastructure. This action highlights the potential misuse of generative AI for transnational influence operations and underscores platform enforcement against security and compliance violations.
OpenAI banned an account linked to a Chinese law enforcement individual for allegedly using AI to plan influence operations, harassment, and cyber activities. This highlights the potential misuse of generative AI in malicious social engineering and information warfare.
OpenAI recently banned accounts originating from Russia that used its models to generate German-language political content regarding Ukraine and NATO. This action highlights the emerging trend of using generative AI for cross-border influence operations, with platforms stepping up efforts to detect and block such misuse.
OpenAI disclosed in its latest threat intelligence report that it banned accounts using ChatGPT to generate bulk political comments targeting the Philippines. Operated by marketing firm Comm&Sense, these accounts posted short texts on TikTok and Facebook to support influence campaigns against President Marcos. While multi-platform, the activity was assessed as Category 2 interference with limited evidence of widespread organic uptake.
OpenAI recently banned a group of likely China-origin accounts that used AI to mass-generate English social posts and Spanish-language articles for influence operations. This action highlights the risks of LLMs being weaponized for automated disinformation and signals stricter platform governance on content authenticity.
OpenAI has banned a set of accounts that used AI to generate articles, posts, and fake engagement during the 2024 Ghanaian presidential election. This action targets covert influence operations leveraging artificial intelligence to manipulate political discourse.
OpenAI confirmed that a viral screenshot claiming to show a Russian troll's expired GPT-4o credits was a hoax orchestrated by US-based accounts. The incident exploited public skepticism about Russian influence operations, leading to the suspension of both the OpenAI and X accounts involved.
OpenAI banned the 'A2Z' account network, which used AI to mass-generate multilingual influence content regarding elections, Ukraine, and geopolitics. This action aims to curb large-scale disinformation and political manipulation via automated means.
OpenAI has banned accounts linked to the Iran-origin 'STORM-2035' influence operation, which used AI to generate content targeting US and UK elections for multi-platform distribution. This action highlights the risks of LLM misuse in political manipulation and the urgency of platform governance.