Russian Disinformation Network Manipulates AI Chatbots, Study Finds

Researchers reveal that Moscow's Pravda network has infiltrated AI training data, causing chatbots to reproduce pro-Kremlin propaganda.

A study by NewsGuard found that 10 leading AI chatbots repeated Russian propaganda 33% of the time, with seven directly citing Pravda network articles as sources.
The Pravda network, a Moscow-based disinformation operation, published 3.6 million articles in 2024 to flood AI training datasets with pro-Kremlin narratives.
Researchers warn of 'LLM grooming,' a tactic where disinformation is deliberately inserted into AI model training data to bias outputs.
False claims, such as Ukrainian President Zelensky banning Truth Social, were repeated by multiple chatbots, illustrating the reach of the Pravda network's influence.
Experts emphasize the need for AI companies to implement stricter safeguards to prevent their systems from amplifying state-sponsored disinformation.