Fabricated audio recordings made to sound like Germany’s Tagesschau are spreading across social media. The clips claim the program is apologizing for alleged lies and accusing demonstrators, but ARD-aktuell editor-in-chief Marcus Bornheim says the recordings are fake.
False apologies travel across platforms
The recordings present supposed apologies over coverage of the war in Ukraine and the coronavirus pandemic. They also include claims about the “denunciation” of demonstrators. The material has circulated through conspiratorial Telegram channels, Facebook and TikTok.
The audio has also moved beyond social media: it was played at a demonstration in Kiel, Germany. That matters because a clip presented in a public setting can reach people who encounter it outside the channels where it first spread. Regardless of where someone hears it, the recording’s claim to represent the news program is false.
Why the recordings can mislead
Bornheim says the voices sound mechanical and use incorrect intonations, signs that listeners may notice. But a familiar program and its recognizable presentation can still make fabricated material seem credible, especially when it appears in a fast-moving social feed.
Bornheim says the audio’s positions and strategies could damage the free and democratic basic order. He also says the Tagesschau’s good reputation is being exploited to spread targeted disinformation. The impersonation therefore does more than attach a false statement to a broadcaster: it borrows the trust people place in the program.
News figures are recurring deepfake targets
The fake Tagesschau audio is part of a broader pattern of impersonation involving news personalities. German anchors Christian Sievers and André Schünke were recently targeted in fraudulent promotional videos on Facebook. The clips used digital copies of the presenters to advertise AI-powered investment platforms and promise high returns.
In those videos, the cloned voices sounded deceptively real, and the synchronized lip movements could appear convincing to people scrolling quickly. The example shows how a fabricated endorsement can be made to look like a familiar journalist is promoting a product, even when the person had no role in the promotion.
News anchors and programs make useful targets for these fakes because audiences know them and there is a large amount of their video and audio available online. That material can be used to train systems that produce imitations. A recognizable voice or face can then lend borrowed credibility to a message the person never delivered.
Verification becomes part of listening
Experts warn that AI-generated fakes can weaken confidence in both individual presenters and media organizations. As the fakes become easier to produce and their quality improves, it may become harder to judge a recording by sound or appearance alone.
That creates a practical challenge for anyone encountering a startling clip: a familiar voice is not proof that a real broadcast took place. In this case, the recordings’ claims are fabricated, and ARD-aktuell has publicly clarified that they are fake. Treating such material cautiously helps prevent a convincing impersonation from being mistaken for a newsroom’s actual reporting.