How ru-propaganda is infiltrating popular AI chatbots
An audit of leading artificial intelligence chatbots revealed that they repeated false narratives “laundered” by the russian disinformation network Pravda in 33% of cases. This is stated in a report by NewsGuard, an American company that analyzes information sources for reliability and the presence of disinformation.
The researchers explained that the “infection” of chatbots occurs in several stages:
▫️ Creation of a fake. For example, a deepfake video is published on YouTube featuring a “whistleblower” who claims an outrageous act of corruption and presents fabricated documents as “evidence.”
▫️ Mass dissemination. The YouTube account that created the fake has few subscribers, but the video statement is picked up by dozens of little-known pro-kremlin websites. At this stage, the Pravda network, whose activities were previously reported by the Center, is actively engaged.
▫️ “Infection.” By massively replicating the false message, ru-resources mislead the chatbots. They cannot identify that the “YouTube whistleblower’s” claims are in fact fake and cite propaganda websites as credible sources. This closes the cycle of legitimizing russian disinformation narratives through AI chatbots.
In this way, the kremlin has opened a “new front” in the information war against Ukraine and the entire civilized world, posing serious challenges to AI developers and users.
