ai chatbots spread propaganda

As technology companies race to develop more advanced AI chatbots, researchers have uncovered a troubling trend: these digital assistants are frequently amplifying Russian propaganda about the war in Ukraine. Multiple independent studies show these AI tools regularly repeat false narratives pushed by the Kremlin.

AI chatbots are inadvertently becoming megaphones for Kremlin propaganda about Ukraine, according to alarming research findings.

A NewsGuard audit found Russian false narratives appeared in about one-third of chatbot answers to Ukraine-related questions. In a separate test, when asked about topics linked to Moscow-based propagandist John Mark Dougan, chatbots shared disinformation in 32 of 57 responses.

The Institute for Strategic Dialogue (ISD) discovered that across 300 Ukraine war questions, chatbots cited Russian state-linked sources in roughly 20% of their answers. The problem was especially bad when discussing Ukrainian military recruitment, where ChatGPT cited Kremlin outlets 28% of the time, while Grok did so in 40% of responses. These AI responses highlight that despite EU sanctions against 27 Russian media entities since February 2022, the sanctioned outlets continue to appear in popular chatbots.

This issue stems largely from what researchers call “data voids” – topics where reliable information is scarce, but propaganda is plentiful. When these voids exist, high-volume Russian content can dominate the information that chatbots access during training or real-time searches.

Russia has developed sophisticated systems to exploit these vulnerabilities. A network called “Pravda” generates thousands of articles across 150 websites in 46 languages to create an artificial consensus around false claims. Many sites pose as local American news outlets, tricking chatbots into treating them as credible sources.

Russian operations also target Wikipedia through coordinated editing campaigns, knowing that many AI systems rely on it as a reference source. The risks are particularly concerning as we approach a period where over one billion people worldwide are set to vote in various elections. These tactics combine with bot networks and search engine optimization to guarantee Kremlin narratives appear prominently in AI training data.

The problem highlights a critical weakness in today’s AI systems. When asked about contested topics like Ukraine, these tools often can’t distinguish between legitimate reporting and state-sponsored disinformation campaigns, inadvertently becoming digital amplifiers for Russian propaganda.

References

You May Also Like

OpenAI Silenced ChatGPT Shooter Warning: Altman’s Belated Regret

OpenAI banned a ChatGPT account for violent activity months before a mass shooting—then decided police didn’t need to know. Altman now apologizes.

AI Pioneer Relieved His Mortality Shields Him From Potential Machine Takeover

AI pioneer Geoffrey Hinton finds comfort in his mortality as he warns younger generations about AI dangers. The former Google scientist now regrets his revolutionary work. The machines we created might outsmart us all.

Tech Publishing Giant Ziff Davis Declares War on OpenAI Over ‘Stolen’ Content

Media giant takes on AI juggernaut as Ziff Davis sues OpenAI for “stealing” thousands of articles. Publishers and AI developers face off in a battle that could reshape digital content laws.

Reddit’s Human Revolution: CEO Defies AI Trend in Bold Content Pledge

Reddit’s CEO defies tech giants by rejecting AI content while competitors embrace automation—but will this gamble destroy the platform?