For many regular social media users, the occasional deepfakes appearing on their feeds might not warrant much attention. Professor Alan Read of King’s College London, for instance, would typically report them or simply scroll past.
This changed when an unfamiliar account tagged him in a video prominently featuring his own face. The synthetic voice, remarkably similar to his own, delivered a politically charged attack on French President Emmanuel Macron. The message referred to him and other Western leaders as being “aboard the Titanic which has ‘European Union’ written on its hull.”
“Almost everything in that video is egregious, and awful to listen to,” Dr. Read, an established theatre professor with no political background, shared with the BBC. “It strikes me as… utterly alien to me.”
Dr. Read’s likeness was used in a new surge of synthetic videos, reportedly linked to Russia, that have circulated widely across social media in recent months. Security experts are raising alarms, suggesting that the West must prepare for a confrontation with the Kremlin on the artificial intelligence front.
“What we’re observing is not merely an increase in deepfakes but a fundamental shift in influence production,” commented Chris Kremidas-Courtney, a defense and security analyst at the European Policy Centre think tank. “We are confronted with systems capable of generating persuasion at scale, at very low cost. This represents a revolution in political influence, and our current governance frameworks are ill-equipped to handle it.”
These videos, some amassing hundreds of thousands of views, have been used to undermine EU institutions and accuse the Ukrainian government of corruption. This occurs precisely as Ukraine seeks financial aid from Western partners to continue defending itself against Russia’s full-scale invasion, now entering its fifth year.
This recent escalation follows several months after the release of Sora2, OpenAI’s latest video generation software, which marked a significant advancement in the realism of its output. In their competition for market share, other AI video tools have sought to attract users by drastically reducing prices or by omitting safety features. One such feature, watermarks applied by Sora2 to differentiate its content from authentic footage, is often absent in these competing applications.
“They need to attract users,” explained Russian AI expert Arman Tuganbaev. He noted that while OpenAI strives to prevent the creation of videos depicting specific individuals, “second-tier apps will offer you that option.” OpenAI itself stated to the BBC that it takes action against accounts involved in deceptive activities intended to cause harm, including making false claims about content origins.
This technological competition has contributed to a consistent rise in both the volume and sophistication of foreign influence operations. This, in turn, strengthens Russia’s position in its ongoing hybrid conflict with Western nations.
In late December, a series of AI-generated videos gained viral traction on TikTok, showing young Polish women advocating for “Polexit,” meaning Poland’s withdrawal from the EU. “There is no doubt that this is Russian disinformation,” stated Adam Szlapka, a spokesperson for the Polish government. “Anyone who looks closely can identify Russian syntax in these videos.” Poland officially requested the European Commission investigate TikTok regarding this incident.
TikTok, which has since removed the clips and the associated accounts, reported taking down over 75 covert influence operations globally in 2025. In the United Kingdom, Members of Parliament have expressed concerns that Russian deepfakes could impact the local elections scheduled for May.
“We have witnessed their extensive use in elections worldwide, so there’s no reason to assume Britain would be an exception,” Vijay Rangarajan, the chief executive of the UK Electoral Commission, informed lawmakers. However, the UK’s Online Safety Act does not explicitly define disinformation as a form of harm. It does, though, require platforms to remove content proven to be foreign influence, a process that is often too slow in an online environment where videos can become viral within hours.
While tracing the origin of these posts can be difficult, Western researchers have identified common characteristics—ranging from stylistic elements to distribution methods—that connect them to organized disinformation units aligned with the Kremlin. One campaign, known as Matryoshka or Operation Overload, is believed to have been responsible for a series of synthetic videos discrediting Moldova’s president, Maia Sandu, during her 2025 election campaign. NewsGuard, an organization monitoring online disinformation, noted recurring patterns suggesting the same network was likely behind the video featuring Dr. Read.
The operation’s name, derived from Russian nesting dolls, reflects its technique: an initial false claim is concealed within layers of reposts from older or compromised social media accounts. Unlike established Russian propaganda outlets such as RT and Sputnik, which Western nations quickly sanctioned at the outset of the invasion of Ukraine, these campaigns offer “a level of… plausible deniability that complicates counter-influence efforts,” according to Sophie Williams-Dunning, a cyber and tech researcher at the Royal United Services Institute think tank.
Researchers at Clemson University have linked a different network, identified by Microsoft’s Threat Analysis Centre as Storm-1516, to former participants of the Kremlin’s “troll factory.” This operation was previously managed by Yevgeny Prigozhin, the deceased leader of the Wagner paramilitary group. In an forthcoming study reviewed by the BBC, these academics provided an example illustrating the rapid spread of fake news on social media.
They observed that each time the Storm-1516 campaign propagated a false narrative—for instance, accusing Volodymyr Zelensky of being “corrupt”—that narrative captured approximately 7.5% of all discussions concerning the Ukrainian president on X within the subsequent week. “That is something any marketing company would be proud of,” remarked Darren L. Linvill, one of the study’s authors.