NewsGuard has issued a warning, indicating that the rate of false information spread by artificial intelligence platforms has almost double
NewsGuard has issued a warning, indicating that the rate of false information spread by artificial intelligence platforms has almost doubled over the past year. These systems are increasingly drawing on a contaminated online information ecosystem, which is sometimes intentionally fed by expansive networks of malicious actors, including Russian disinformation campaigns.
With the integration of real-time internet search, chatbots have ceased their previous refusals to answer questions. According to NewsGuard’s report, published in early September: “Instead of citing data limitations or declining to comment on sensitive topics, LLMs now draw from a polluted online information ecosystem – sometimes deliberately fuelled by extensive networks of malicious entities, including Russian disinformation operations – and treat unreliable sources as credible.”
The study examined 10 leading AI tools and their tendency to repeat false claims about news-related topics, finding that they repeated false information in over one-third of cases.
In August 2025, the rate reached 35%, compared to 18% in August 2024. The report concludes: “When it comes to providing reliable information about current events, the industry’s promises of safer and more trustworthy systems have not translated into actual progress.”
One example cited in the report, titled “AI Misinformation: Rate Nearly Doubled in a Year,” concerns widely circulated fake information about President Zelensky’s supposed multibillion-dollar real estate empire. The case implicates Mistral, a Paris-based company that has played a key role in strengthening Europe’s position as a leader in responsible AI development.
When asked about the alleged properties, Mistral confirmed their existence, citing an article from the Kremlin-linked network Pravda as its source. The operator of Pravda, Yevgeny Shevchenko, and his IT company TigerWeb were sanctioned by the European Union in July 2025 for spreading disinformation In other words, Mistral confidently repeated a false claim propagated by Russia and directly referenced a disinformation source subject to sanctions.
COMMENTS