Russian Disinformation 'Infects' Popular AI Chatbots
A Russia-based disinformation network has successfully "infected" many of the world’s most popular AI chatbots with pro-Kremlin misinformation, according to a new report by NewsGuard.
AI Grooming and Pro-Kremlin Network
Rather than targeting readers with propaganda directly, the network reportedly publishes millions of articles in different languages, pushing its narratives across the web, hoping they will be incorporated as training data used by large language models (LLMs) like OpenAI’s ChatGPT or xAI’s Grok. This practice has been dubbed "AI grooming" by NewsGuard.
Impact on AI Chatbots
NewsGuard audited 10 of the most popular AI chatbots, including OpenAI’s ChatGPT-4o, You.com’s Smart Assistant, xAI’s Grok, and others, querying them about 15 pro-Russia narratives spread by the Pravda network. The findings revealed that some chatbots regurgitated false claims sourced from the disinformation network.
For instance, some chatbots cited misinformation such as members of the Ukrainian Azov Battalion burning effigies of President Trump and other fabricated stories.
Volume of Disinformation
Despite having minimal organic reach, the Pravda network managed to effectively influence mainstream chatbots by saturating search results with a substantial volume of content. The network publishes an average of 20,273 articles every 48 hours, totaling roughly 3.6 million articles per year.
Russian Influence on AI Models
The impact of Russian disinformation varied across different chatbots, with some showing a higher susceptibility to false narratives than others. Russian President Vladimir Putin has openly discussed the importance of controlling the narratives of AI models and search engines, highlighting concerns about bias in Western AI standards.
AI and Propaganda
Online Russian disinformation tactics have evolved to utilize AI in propagating false narratives. OpenAI has identified Chinese linked accounts using ChatGPT to create propaganda articles from scratch for mainstream Latin American newspapers.
Legitimate publications have taken measures to prevent LLMs from scraping their content, with some resorting to legal action and others forming partnerships with AI companies to regulate the use of their information.
As AI continues to play a significant role in information dissemination, the battle against misinformation remains ongoing.
