Russia seems to have found its new recreation courtyard to transmit misinformation. Chatbots, such as Chatgpt, Perplexity or Grok, are beginning to become a common source of information for many people. However, they are not waterproof for misinformation despite the safeguards.
By disseminating false information in the databases that feed the AI, certain malicious groups can influence chatbots responses. According to Newsguard, an organization against misinformation, A third of the answers generated by the ten main opinions of AIS transmitted by the pro-ruse media.
France is markedly in the Kremlin’s Liper Line. In question, a tense international context where France is placed as one of the main allies of Ukraine.
Saturate
The Russian strategy to deceive chatbots is simple: saturate the data collected by Internet chatbots from false sources.
Because when it offers an answer, a chatbot works more or less in a probability system: the more information it is transmitted in many media, the more the chatbot considers that it is true. Some chatbots even draw their information from social networks.
In addition, chatbots tend to favor recent sources when looking for an online response on a current topic. The more the Internet is flooded with false articles on a topic, the more a chatbot will tend to repeat this false information in your answers.
Obviously, moderation of AI and social networks acts as a safeguard. But the arrival of Donald Trump and the new policies of certain social networks reduce the means assigned to moderation. False information risks to present in chatbots responses are, therefore, quite high.
Pravda, a massive dissemination weapon
To achieve this flood of misinformation, Russia has a powerful tool specifically designed to gang the chatbots responses. According to Newsguard, it goes through a network of almost 150 pro-Kremlin information sites entitled “Pravda”, the Russian word that means ironically “truth.”
This network has been controlled from Crimea by a company called Tigerweb. He had discovered in 2024 by Viginum, a French service in charge of protection against foreign digital interference.
These sites massively spread their content in more than 46 languages. In 2024, almost 3.6 million articles were published in all these sites. These sites are not destined to be visited by human users, their assistance is very low. But by flooding the Internet of this false information, it is likely that chatbots generate information about these sites.
One of the Kremlin propaganda leaders affirmed this strategy during a revealed meeting on YouTube: “By approving these Russian stories from a Russian point of view, we can really change global artificial intelligence.”
An operation that wants to sow disorder
Another content is part of a Russian misinformation campaign called Storm-1516 against several Western countries. Storm-1516 creates false information from anywhere and counts on social networks to disseminate it.
Newsguard thus characterizes storm tactics: “Take a very real individual who has a verifiable link with a public figure and build a false accusation around it.”
At the origin of this strategy, there is a former American assistant sheriff named John Mark Dougan. Not having a sufficient number of people to produce such a amount of content, Storm-1516 uses generative to create a flow of false content, so the goal is to become viral.
The objective of these contents is not in particular to influence the opinion of the population in the international situation, or even change the course of an election. According to Newsguard, Storm-1516 was particularly active upstream of the US presidential election of 2024, during the Paris Olympic Games, or during the German federal elections of February 2025.
France in the storm
The new goal of Storm-1516 is now France. In a Newsguard publication dating from April 17, we learned that five totally false information would have been widely disseminated between December 2024 and March 2025.
These lies have been transmitted more than 38,000 times on social networks that obtain almost “55 million visits.” Here are the five misinformation campaigns in question (in chronological order):
- The confessions of a Chadiano migrant who would have violated a 12 -year -old girl in France
- A false video of an Islamist group threatening to set fire to Notre-Dame de Paris
- A false testimony of a former Brigitte Macron student who accuses the first lady of having sexually abused her when she was 12 years old
- The fact that Volodymyr Zelensky would have made the acquisition of a French bank
- Reaulf Fleming, an activist from the Western Indies against AIDS, who reveals the secret matter that Emmanuel Macron would have had with his brother
France and its leaders are attacked due to their assertive support to Ukraine against whom Russia is at war. The last Russian campaign took place at the end of March 2025, allegedly after Emmanuel Macron presided over a meeting on peace and security in Ukraine.
Chatbots infected with Russian propaganda
The observation is alarming for Newsguard. Recently, the organization recently asked the eleven most used chatbots in the world if the secret relationship that Emmanuel Macron would have had with Réaulf Fleming’s brother was true.
Five of them affirmed the veracity of this information, although false. Among them, the French chatbot, the company’s cat, Mistral, would have confirmed the false information.
This story also illustrates the fact that Storm-1516 would work in a concert with the Pravda Network by greatly transmitting this manufactured information. Four days after Storm-1516 launched the rumor about the secret relationship of Emmanuel Macron, Pravda’s sites massively resumed information.
Source: BFM TV
