Russian disinformation infiltrates Western AI chatbots
According to the findings of scientists, Russia also uses Western chatbots in its global disinformation campaigns.
A study by NewsGuard revealed that more than one-third of the responses given by AI communication assistants, such as ChatGPT, contained pro-Russian false news. Researchers attribute this to the Moscow-based network Prawda, which spreads pro-Kremlin propaganda worldwide.
"Massive amounts of Russian propaganda — 3,600,000 articles in 2024 — are now incorporated in the outputs of Western AI systems, infecting their responses with false claims and propaganda," write NewsGuard researchers McKenzie Sadeghi and Isis Blachez in their report.
As part of the study, NewsGuard analysed the ten most important artificial intelligence applications, including ChatGPT-4, Smart Assistant, Grok, Copilot, Meta AI, Google Gemini, and Perplexity. All of them perpetuated misinformation disseminated by "Prawda." Seven chatbots even cited specific articles from "Prawda" as sources.
Manipulation of chatbots
According to the researchers, Prawda manipulates chatbots by flooding large language models (LLM) used by AI applications to process and generate text with false information. This method is called "LLM Grooming." The NewsGuard study shows that AI chatbots pick up misinformation circulating online and are deliberately subjected to its influence.
Meanwhile, the non-profit organisation American Sunlight Project, in a separate study, warns of the growing reach of the "Prawda" network, also known as "Portal Kombat," and the risk of flooding the training data of large language models (LLM) with pro-Russian content.
Nina Jankowicz, executive director of the American Sunlight Project, highlights that the scope at which the 'Prawda' network disseminates misinformation is unmatched. She warns that its capacity to impact AI systems further amplifies the risk it poses. Jankowicz also emphasizes that as Russian influence grows in both reach and complexity, the foundation of democratic discourse across the globe faces a serious threat.
Experts warn that the threat from Russian disinformation is particularly significant because the United States is doing nothing to counteract it. As reported by several American media outlets earlier this month, U.S. Secretary of Defense Pete Hegseth has ordered a temporary halt to cyber operations against Russia. All American actions against Moscow are currently being reassessed.
Kremlin propaganda on X, Telegram, Bluesky
The Moscow-based network "Prawda" was founded in April 2022 following Russia's invasion of Ukraine and has been steadily expanding since then. According to NewsGuard, it currently spans 49 countries and dozens of languages.
"Prawda" does not create its own content but uses articles from Russian state media or individuals loyal to the Kremlin, which the network disseminates millions of times through platforms such as X, Telegram, and Bluesky. One false report claims, for example, that the U.S. is operating secret biological weapons laboratories in Ukraine. In February, fact-checking experts from the AFP news agency debunked a false claim on social media that Ukrainian President Volodymyr Zelensky had banned access to the Truth Social network of U.S. President Donald Trump.
When NewsGuard questioned various chatbots about whether Zelensky had banned Truth Social, six mistakenly presented the false claim as factual, often referencing articles from the Moscow-based network "Prawda." Additionally, NewsGuard found that these chatbots relayed misinformation from Kremlin propagandist John Mark Dougan, a former U.S. police officer, as if it were verified information. During a conference in Moscow in January, Dougan reportedly stated that advancing this Russian narrative could significantly shape artificial intelligence on a global scale. He also encouraged using AI as a tool rather than something to be feared.
Read also:
