Audio clip of President Joe Biden talking about bank collapse is fake, experts said

An audio clip is making the rounds on TikTok and Twitter, where social media users are claiming it is a recording of President Joe Biden discussing the fallout from the recent Silicon Valley Bank collapse. 

In one TikTok video, posted March 15, viewers see an image of a vintage tape recorder, and hear a voice similar to Biden’s. "It looks like all the money is gone, and a lapse — a flaps — flapper, uh, a collapse is imminent," the voice says. "But we have to uh, we have to, uh, we have to stay calm and think of ways to vent, uh, vent, vent, um, prevent a bank run." The text on the video says, "Biden on tape talking about us not panicking when the collapse happens. I warned you guys a year ago." 

The voice in the audio clip also says "Let’s use the full force of the media to calm the public." 

Comments on the post showed that some people believed the audio was real, while others knew it was fake. 

"I don’t want to believe this because it could have been spliced together. But it seems to b more than happening, so guess its time to (m)ake a withdrawal!" wrote one. Another person commented, "Just so everyone knows this has confirmed to be fake."

The second commenter was correct: The audio is fake. TikTok identified this video as part of its efforts to counter inauthentic, misleading or false content. (Read more about PolitiFact's partnership with TikTok.) 

White House spokesperson Michael Kikukawa told PolitiFact that the audio "is inauthentic." 

An expert said clues in the audio prove it’s not real.

Hafiz Malik, a University of Michigan, Dearborn, professor who researches multimedia forensics and deepfake technology, analyzed the audio clip forensically and said the clip was machine-generated and likely created with a low-grade audio tool.

"When I did my analysis, it had surprisingly low frequency, which typically in human voice, you don’t see," Malik said. 

Featured Fact-check

Facebook posts

stated on December 1, 2024 in a Facebook post

“Chick fil-A’s new policies have people thinking (of a) boycott.”

According to Flypaper, a magazine from music education platform SoundFly, the frequencies of adult human voices range from 80 to 255 hertz during a conversation, varying for males and females. When Malik analyzed the audio clip, the frequency was 10 hertz. 

"So that is basically not the human vocal cord doing something," Malik said. 

Social media users can suss out fake audio on their own, Malik said. For example, the audio that claimed to be of Biden was missing a distinct breath between words — something that Biden does when talking, Malik said.

"If you listen to his audio, you see that hissing sound, like he has to inhale air basically before he starts speaking. That’s pretty evident if you listen," he said. "There’s nothing like that in this (TikTok) recording."

Mike Caulfield, a research scientist at the University of Washington’s Center for an Informed Public, shared more pointers. To begin, he said, people should find out where the audio originated. 

"As (artificial intelligence) production of media accelerates, the best way for the average person to verify audio is to examine provenance, not surface features of the video or audio," Caulfield said. "Where did the audio come from? Why does this particular person have access to it? Where was it supposedly recorded, and in what context?"

The second to check is whether reputable sources can verify the audio’s authenticity. 

"What we encourage people to do is take some segment of what is said in the audio and search for it. Sometimes this will surface the words said by the person — but in a different context. Sometimes it will surface a fact-check, or original reporting. In this case, it surfaces only TikTok results, which is a pretty good indication it can be ignored," Caulfield said.

A TikTok video claims Biden was recorded talking about the Silicon Valley Bank collapse.

The White House said the audio clip is inauthentic and an expert said it’s fake. The expert’s forensic analysis showed that the voice was machine-generated. 

We rate this claim Pants on Fire!