A Russian-based disinformation network has successfully infected many of the most popular Chatbots in the world with Pro-Cremlin misinformation, according to a new report by Newsgard.
Instead of targeting readers with direct propaganda, the network is reported to publish millions of articles in different languages, pushing its narratives across the network, hoping that they will be included as training data from large language models such as Chatgpt Openai or Grok of X. Newsgard called this practice “he grooming”.
The pro-Mremlin network, known as Pravda, which is Russian for the truth, began shortly after the Russian occupation of Ukraine in April 2022 and has been gradually growing on a scale to about 150 pages on the Internet.
Newsguard Audit 10 of the most popular chatbots of him: Openai’s chatgt-4o, smart assistant of Ju.com, Grok I XAI, Pi Inflection, Le chat of Mistral’s, Microsoft Copilot, Meta AI, Claude Anthropic’s, Google Bemini, and ENGINE OF AFTERS. Newsguard asked chatbots about 15 pro-Russian narratives that have been advanced by a network of Pravda’s websites since the beginning of the war.
For example, Newsguard claims that four of the 10 chatbots that were appreciated with regulated allegations that members of the Ukraine Azov Battalion burned President Trump’s efficiency, citing articles from the misinformation network as their resources.
Other false claims, the Pravda network spread that Newsguard used in the analysis included the French police, saying that an official from the Zelensky Ministry of Defense stole $ 46 million and that Zelensky personally exceeded 14.2m euros in Western military funds to buy a famous German retreat by Adolf Hitler.
The disinformation network managed to effectively affect many of these main chatbots with barely any organic extension. Pravda-en.com, an English language site within the network, only on average 955 unique monthly visitors.
However, the operation focused on saturating the search results with a large volume of content. The report from the US Sun Project (ASP) found that, on average, the network publishes 20,227 articles every 48 hours, or approximately 3.6 million per year.
But the impact of Russian misinformation varied widely depending on what Chatbot researchers viewed. One chatbot cited 55% of the time after introducing it to fake narratives, while another did it just over 6% of the time. (Newsguard did not find out which particular chatbot was behind each result.)
Recommended by our editors
The highest levels of the Russian leadership have already openly discussed the importance of controlling the narratives of AI and the search engines.
Russian President Vladimir Putin said in a 2023 conference that he “created in accordance with Western standards and models could be xenophobic” and that “Western research engines and generating models often work in a very selective, unilateral way”.
Russian online misinformation is nothing new, but it is being used in increasingly creative ways for propaganda. Openai has highlighted Chinese related accounts using chatgpt to produce propaganda articles from scratch for publication in Latin American newspapers.
Get our best stories!
register What’s new now To get our main stories delivered to your box every morning.
This bulletin may contain advertising, agreement or cooperation links. By clicking the button, you confirm that you are 16+ and agree with our use conditions and privacy policy. You can subscribe from newspapers at any time.
About Will McCurdy
contributor
