- A new Institute for Strategic Dialogue (ISD) report reveals that OpenAI’s ChatGPT, Google’s Gemini, DeepSeek, and xAI’s Grok frequently cite sanctioned Russian state media (e.g., RT, Sputnik, EADaily) and pro-Kremlin sources in responses to queries about Russia’s war in Ukraine, with nearly 18 percent of outputs across 300 tested questions in five languages linking to such entities.
- Researchers exploited “data voids” in real-time information gaps to test neutral, biased, and malicious prompts on topics like NATO perceptions, peace talks, Ukrainian recruitment, refugees, and war crimes. Malicious queries yielded pro-Russian content 25 percent of the time, compared to 10 percent for neutral ones, showing confirmation bias in LLMs.
- The EU has sanctioned at least 27 Russian media outlets since 2022 for disinformation, including recent 2025 bans on eight more (e.g., Fondsk, NewsFront, SouthFront). Chatbots’ amplification raises regulatory concerns under the Digital Services Act, as ChatGPT nears Very Large Online Platform status with 120M+ EU users.
- Russian networks like “Pravda” flood the web with 3.6M+ AI-generated articles yearly to “poison” LLM training data, per NewsGuard; OpenAI attributes issues to search integrations, while xAI dismissed claims as “Legacy Media Lies,” highlighting foreign influence risks as AI replaces traditional search. Read More
Menu
