Leading AI chatbots, including ChatGPT, Google Gemini, and Meta AI, are generating and disseminating Russian disinformation narratives, according to the findings of a NewsGuard study published by Axios on June 18.
NewsGuard is a digital transparency company that aims to counter misinformation in online media. For this study, analysts entered prompts related to known Russian disinformation campaigns to test the responses of 10 popular chatbots.
The study found that chatbots responded with Russian propaganda 32% of the time, citing fake news sites as if they were credible sources.
"What's really alarming is that hoaxes and propaganda these chatbots repeated so frequently were hardly obscure," NewsGuard co-CEO Steven Brill told Axios.
The study involved prompts related to narratives disseminated by John Mark Dougan, a former U.S. police officer who now spreads Kremlin propaganda in Moscow. The chatbots would then frequently regurgitate Dougan's claims, presenting them as fact.
Anna Myroniuk
Among the chatbots' false claims were fake news about a Ukrainian troll factory interfering with U.S. elections and a wiretap at former U.S. President and presumptive Republican nominee Donald Trump's Mar-a-Lago residence. Dougan's propaganda sites were cited as legitimate news sources in these cases.
"This report really demonstrates in specifics why the industry has to give special attention to news and information," Brill said.
Analysts conducted their study using OpenAI's ChatGPT-4, You.com's Smart Assistant, Grok, Inflection, Mistral, Microsoft's Copilot, Meta AI, Anthropic's Claude, Google Gemini and Perplexity.
NewsGuard contacted the companies to seek comment on the results, but did not receive any responses.
"(D)on't trust answers provided by most of these chatbots to issues related to news, especially controversial issues," Brill said.
Liliane Bivings
