The Department of Justice says it will stop AI-generated disinformation before it starts by concealing an audio recording of President Joe Biden
The Department of Justice (DOJ) refused to release an audio recording of President Biden’s interview with former special counsel Robert Hur in a court filing on Friday, obtained by Politico. The department cited fears that publicizing the audio could lead to deceptive AI deepfakes of the sitting President.
“For example, a malicious actor could slow down the speed of the recording or insert words that President Biden did not say or delete words that he did say,” the department said in the filing. “That problem is exacerbated by the fact that there is now widely available technology that can be used to create entirely different audio ‘deepfakes’ based on a recording.”
More than 13 media organizations and several conservative groups have filed requests to publicize the audio recording under the Freedom of Information Act (FOIA), according to the DOJ filing. A transcript of the interview, which pertains to Hur’s investigation into Biden’s alleged unauthorized removal and retention of classified documents, was already made public in March. However, Politico reports that those seeking the audio would like to hear Biden’s tone and pauses, while some House Republicans speculate the transcript may not perfectly match the audio.
The fight for the Biden-Hur interview audio has outlived the investigation. Hur’s 345-page report concluded in February that no charges should be brought against the President, but that Biden was a “well-meaning, elderly man with a poor memory.” Republicans may want to obtain this audio to reinforce this image of Biden, but some argued it could provide insights into their impeachment inquiry. Just days after the transcripts were released a month later, Biden invoked executive privilege to block House Republicans from obtaining the audio, according to The Hill.
This appears to be the first time the United States has rejected to release recordings of a sitting President on the grounds of artificial intelligence concerns. The problem is that there’s already tons of footage and recordings of Biden speaking. In January, some bad actors utilized AI to create robocalls that sounded like President Biden. These calls were sent to New Hampshire voters and made it seem like President Biden was telling them not to show up for a primary election.
The DOJ even noted that it’s still possible to create deepfakes of Biden without this interview recording. That said, the department says releasing the audio would make it more difficult to quash disinformation.
“To be sure, other raw material to create a deepfake of President Biden’s voice is already available, but release of the audio recording presents unique risks: if it were public knowledge that the audio recording has been released, it becomes easier for malicious actors to pass off an altered file as the true recording,” said the filing.
The DOJ is effectively saying it’s easier to stop AI-generated disinformation if we release nothing at all, which seems like a dangerous precedent to set. While this argument is true in some ways, it reflects the sensitive nature of these interviews. President Biden speaks all the time, but these other recordings aren’t granted the same level of security.
Join 500,000+ readers who start their day with Quartz.
By subscribing, you agree to our Terms of Service and Privacy Policy.
