One third of responses generated by conversational AIs on topics related to the war in Ukraine or international politics are reportedly false and influenced by Kremlin propaganda.
Several recent investigations reveal that content from the “Pravda” network (also known as “Portal Kombat”), a group of 150 propaganda websites publishing in 46 languages, has been incorporated into the training data of certain AIs such as ChatGPT, Gemini, and Le Chat.
This propaganda network produces 1,500 articles per day in multiple languages to manipulate public opinion, promote Kremlin interests, and attack Ukraine and the West. The more frequently a piece of information is repeated across platforms and in different languages, the more credible it appears to AI algorithms — and the more likely it is to be relayed.
Among the false narratives spread through these tools is, for instance, the claim that Ukrainian President Volodymyr Zelensky banned Truth Social, Donald Trump’s social media network, in Ukraine.
Such fake news is designed to sow confusion and polarize opinions. When widely disseminated, it gradually shapes public perception and reinforces disinformation.
As millions of users worldwide rely on conversational chatbots every day, combating disinformation has become a critical challenge in the information warfare waged by the Kremlin.
Sources : France 24, franceinfo, Le Monde
