Russian Storm-1516 network uses DeepSeek AI for European influence operations

The use of synthetic media by Storm-1516 demonstrates that current platform regulations like the Digital Services Act are easily circumvented by moving content to messaging apps like Telegram. While major platforms face heavy fines, the migration of AI-generated propaganda to unregulated services and non-EU candidate countries creates a fragmented information environment that is difficult to police. This shift forces a strategic pivot from technical moderation to investing in independent journalism and civic media literacy to counter automated narratives. Watch for the European Commission to potentially expand the list of very large online platforms to include Telegram to close these enforcement loopholes.

Storm-1516 has evolved into one of the most prolific Russian influence networks targeting European audiences since its emergence in 2023. The operation, which the EU's East StratCom Task Force has tracked since its early days, generated more than 274 million views across 34 distinct false narratives about Ukraine between January 2025 and mid-2026, according to the Center for European Policy Analysis. The network's shift to DeepSeek models represents a broader trend of state-aligned actors adopting cheaper, less restricted AI tools to scale content production while evading detection systems trained on outputs from Western models like OpenAI's GPT series.

The European Commission's enforcement of the Digital Services Act has focused primarily on very large online platforms, but critics argue the framework leaves significant gaps. In December 2024, the Commission opened formal proceedings against X under the DSA for failures in content moderation and transparency, marking one of the first major enforcement actions under the regulation. However, Telegram, which has become a primary distribution channel for Storm-1516 content, has not yet been designated a very large online platform under the DSA, meaning it faces lighter obligations. The Hungarian political dimension adds further complexity: Meta content moderation failure in Hungary has accused Prime Minister Viktor Orbán's government of amplifying Russian-aligned narratives through state media, illustrating how influence operations exploit domestic political fractures alongside cross-border disinformation.

Technical analysis of AI-generated propaganda reveals growing sophistication in evading platform detection. Researchers at the Atlantic Council's Digital Forensic Research Lab documented that DeepSeek-generated content can bypass watermarking and provenance systems that flag outputs from OpenAI and Anthropic models, making automated moderation significantly harder. The DFRL found that Storm-1516 operators used DeepSeek to produce multilingual articles in Hungarian, German, and French, then distributed them through coordinated inauthentic behavior on X and Telegram simultaneously. This multi-platform approach, combined with the lower cost of DeepSeek API access compared to Western alternatives, allows the network to produce content at scale while maintaining plausible deniability about AI involvement. The findings underscore why regulators are considering whether to extend DSA obligations to messaging platforms and whether should be mandated for content distributed at scale.