BRUSSELS — European officials warned on Thursday that pro-Russian groups are systematically seeding popular AI chatbots with propaganda, aiming to influence public opinion through the tools many people now use for information.
Senior officials across the European Union expressed concern over efforts by networks linked to the Kremlin that flood the internet with low-quality content designed to shape responses from large language models. The warnings come as AI chatbots gain popularity as sources of news and analysis ahead of key elections.
French officials and other EU member states have highlighted operations such as the so-called Pravda network or Portal Kombat, which consists of numerous websites producing pro-Russian narratives in multiple languages. These sites, identified in earlier reports by France’s Viginum agency, generate content primarily to influence AI training data rather than attract human readers, officials said.
The tactic, sometimes referred to as “LLM grooming,” involves creating data voids or flooding online sources with consistent messaging on topics including the conflict in Ukraine. Researchers have documented cases where major commercial chatbots repeated Kremlin-aligned claims, according to reports from organizations monitoring disinformation.
“Democracies must act together or risk letting AI quietly shape what voters believe,” one senior European official told NBC News in a report published on June 26.
The issue gained renewed attention following studies showing that chatbots from providers including OpenAI, Google and others have cited sources tied to Russian state media or affiliated networks in responses about geopolitical events. In some tests, the share of problematic responses increased over time, though exact figures vary by model and query.
European authorities have coordinated efforts through bodies like the European External Action Service and national agencies to counter foreign information manipulation and interference. Officials stressed the need for collaboration with technology companies to improve safeguards in AI systems.
Russia has not publicly commented on the specific allegations in recent days. Moscow has previously denied engaging in disinformation campaigns, describing such accusations as politically motivated.
The development forms part of broader concerns about hybrid threats in the context of Russia’s ongoing conflict with Ukraine. Details on the scale of impact on public discourse remain under assessment, officials said.
As of Friday, no new coordinated EU-wide measures were announced in direct response to the latest warnings. Technology companies have in the past adjusted models or added disclaimers in response to similar concerns, but the effectiveness of such steps continues to be evaluated by researchers and regulators.
The next steps are expected to involve further discussions among EU member states and engagement with AI developers on data sourcing and model transparency.
