AI Chatbots Found Spreading Russian Disinformation About Ukraine War, Study Shows

Russia-linked content appearing in AI chatbot responses has raised serious concerns about disinformation risks, according to a new analysis from the British think tank Institute for Strategic Dialogue (ISD).

Researchers examined how four leading AI platforms—ChatGPT, Gemini, Grok, and DeepSeek—responded to questions about Russia’s invasion of Ukraine across five languages: English, Spanish, French, German, and Italian. Their findings reveal potentially troubling patterns in how artificial intelligence systems handle contentious geopolitical information.

Nearly one-fifth of all responses referenced Russian state sources, many of which are currently subject to European Union sanctions. The study found that questions framed with a pro-Russian bias were more likely to yield answers containing these sanctioned sources.

Particularly problematic areas included AI responses to queries about Ukraine’s military conscription policies and NATO’s role in the conflict. The ISD noted that the chatbots struggled most when trying to identify Russian state-affiliated content that had been republished or laundered through third-party websites or media outlets.

“The challenge of identifying and restricting sanctioned media sources isn’t new to tech companies,” an ISD spokesperson explained. “Google has faced similar scrutiny over search results related to complex topics, particularly since Russia’s full-scale invasion began and the EU imposed sanctions requiring restriction of state media content.”

This investigation raises fundamental questions about AI systems’ ability to comply with EU sanctions against Russian state media. As generative AI becomes increasingly integrated into information ecosystems, its vulnerability to manipulation poses growing concerns for policymakers, tech companies, and users alike.

The findings come amid broader warnings about Russia’s evolving information warfare capabilities. Intelligence agencies have documented Russia’s intensified efforts to influence American public opinion through an ecosystem of disinformation tactics, including deepfakes, fabricated websites, and now, potentially, manipulating AI systems to amplify Kremlin narratives.

Tech industry analysts point out that AI systems are only as reliable as their training data. If that data includes significant volumes of Russian state media or content influenced by it, those perspectives may be inadvertently reinforced in AI outputs.

“We’re seeing the next frontier of information warfare,” said Dr. Elena Mikhailova, a disinformation researcher not affiliated with the study. “Rather than simply creating fake content, hostile actors can now potentially engineer questions that extract biased or misleading information from widely used AI tools that many people trust implicitly.”

The European Commission has already signaled concerns about AI’s potential role in spreading disinformation. Internal Market Commissioner Thierry Breton previously warned tech companies about their obligations under the Digital Services Act to mitigate risks related to generative AI.

Ukrainian President Volodymyr Zelenskyy has called for restrictions on AI technology exports to Russia, recognizing the dual-use potential of these systems for both civilian and military applications. His administration has emphasized that Russia is rapidly incorporating AI capabilities into its information operations against Ukraine and Western democracies.

For everyday users, experts recommend approaching AI-generated content with healthy skepticism, particularly on contentious geopolitical topics. Cross-checking information with multiple reliable sources remains essential.

Tech companies have responded to the ISD report by reiterating their commitment to addressing these challenges, though specific remediation plans vary by platform. Several have promised to strengthen content filters and improve detection of sanctioned sources.

As AI systems continue to evolve and integrate more deeply into our information landscape, the battle against their potential weaponization in geopolitical conflicts presents a growing challenge for democratic societies worldwide.