AI voice generation is no longer only a tool for creators, product teams, or synthetic media experiments. According to Recorded Future, it has also appeared in a Russian-tied influence campaign aimed at European audiences.
The Massachusetts-based threat intelligence company said commercial AI voice generation products, including technology publicly released by ElevenLabs, were “very likely” used in videos tied to “Operation Undercut.” The campaign sought to erode Europe’s support for Ukraine through fake or misleading “news” clips.
What Recorded Future Found
Recorded Future described “Operation Undercut” as a Russian-tied campaign built around videos that looked or sounded like news content. The videos targeted European audiences and pushed themes that attacked Ukrainian politicians as corrupt or questioned whether military aid to Ukraine was useful.
One example cited in the source article was a video claiming that “even jammers can’t save American Abrams tanks,” a reference to devices used by US tanks to deflect incoming missiles. The message behind that clip was direct: sending advanced armor to Ukraine was being framed as pointless.
The key technology angle was the apparent use of synthetic narration. Recorded Future said the creators “very likely” used voice-generated AI, including ElevenLabs technology, to make the videos seem more legitimate.
That matters because voice is a trust signal. A polished narrator speaking the audience’s language can make low-quality or deceptive material feel more professional, especially when the content is packaged as news.
How ElevenLabs Was Linked
Recorded Future’s researchers tested the audio by submitting clips to ElevenLabs’ own AI Speech Classifier. The tool is described as giving anyone the ability to “detect whether an audio clip was created using ElevenLabs.” According to the source article, the researchers got a match.
ElevenLabs did not respond to requests for comment. Recorded Future also noted the likely use of several commercial AI voice generation tools, but it did not name any others besides ElevenLabs.
The source article also points to an accidental comparison that helps explain why synthetic voiceovers can be attractive in influence work. Some videos were released with real human voiceovers that had “a discernible Russian accent.” By contrast, the AI-generated voiceovers spoke in multiple European languages such as English, French, German, and Polish without foreign-sounding accents.
That difference is important. A campaign trying to reach audiences across Europe benefits when narration sounds fluent, local, and consistent. Recorded Future said AI also allowed misleading clips to be quickly released in multiple languages spoken in Europe, including English, German, French, Polish, and Turkish.
The Campaign Behind The Clips
Recorded Future attributed the activity to the Social Design Agency, a Russia-based organization. The U.S. government sanctioned the organization this March for running “ a network of over 60 websites that impersonated genuine news organizations in Europe, then used bogus social media accounts to amplify the misleading content of the spoofed websites.”
According to the U.S. State Department language included in the source article, the activity was done “on behalf of the Government of the Russian Federation.”
The campaign’s goal, as described by Recorded Future, was not simply to publish isolated videos. It was to support a broader narrative environment in which Ukraine looked corrupt, military aid looked ineffective, and European support appeared less worthwhile.
Still, the reported impact was limited. Recorded Future concluded that the overall effect of the campaign on public opinion in Europe was minimal.
Why Synthetic Voice Changes The Risk
The facts in the source article point to a practical reason AI voice generation is useful in influence operations: scale. A campaign can produce narration in several languages without relying on a large group of fluent speakers.
It also points to another advantage: presentation. A fake or misleading video can look more credible when it has a clean, confident voiceover. If the audio lacks an obvious foreign accent, viewers may be less likely to notice that the message is coming from an outside operation.
That does not mean the technology alone determines whether a campaign succeeds. In this case, Recorded Future found the public opinion impact in Europe was minimal. But the use of AI voice generation shows how commercial tools can lower the production burden for misleading content.
The source article also places the ElevenLabs finding within a broader pattern of alleged misuse. Bloomberg reported that ElevenLabs technology was behind a robocall impersonating President Joe Biden that urged voters not to go out and vote during a primary election in January 2024, according to a voice fraud detection company.
In response to that incident, ElevenLabs said it released new safety features, including automatically blocking voices of politicians. The company also bans “unauthorized, harmful, or deceptive impersonation” and says it uses tools such as automated and human moderation to enforce its rules.
The Larger Stakes For AI Voice Tools
ElevenLabs has grown quickly since its founding in 2022. TechCrunch previously reported that the company grew ARR to $80 million from $25 million less than a year earlier and may soon be valued at $3 billion. Its investors include Andreessen Horowitz and former Github CEO Nat Friedman.
That growth shows why the issue is bigger than one campaign. AI voice generation products are becoming more capable, more visible, and more widely available. When they are used for legitimate work, they can speed up production. When they are used deceptively, they can help influence operators create more convincing content faster.
The Recorded Future report, as summarized in the source article, does not claim that “Operation Undercut” changed public opinion in Europe in a major way. Its conclusion was the opposite: the impact was minimal. But the case still gives a clear warning about the next stage of synthetic media misuse.
For audiences, the lesson is straightforward. A fluent voiceover, a news-like format, and a familiar language do not prove that a video is trustworthy. For AI companies, the challenge is equally direct: tools built for realistic speech can also be pulled into deceptive political messaging, even when policies ban that behavior.