The US Department of Justice (DOJ) has refused to release audio of President Joe Biden’s special counsel interview, citing concerns that the recording could be subject to artificial intelligence (AI) disinformation, including deepfakes. The DOJ has weighed the risks and decided against releasing the audio, fearing it could be maliciously altered and circulated as genuine. Associate Deputy Attorney General Bradley Weinsheimer believes that malicious actors might insert or delete words, reshaping the narrative consistency.
Despite the DOJ concluding that there were no grounds to prosecute Biden, the transcript of the interview has provided critics substantial grounds to question his mental capacities and his ability to compete for another term. Notably, AI-generated voice replicas require only a few seconds of the original voice to fabricate convincing audio clips. An instance was seen in January, where a cloned voice of Biden was used in a false robocall campaign discouraging potential voters.
Even though creating a deepfake of Biden’s voice is possible, the DOJ insisted that releasing the audio recording presents a unique threat. If the release of the audio was public knowledge, it would be simpler for miscreants to circulate an altered file as the original recording.
Although the concerns about deepfake risks seem legitimate, Executive Director of the Oversight Project at the Heritage Foundation, Mike Howell, accuses the DOJ of using fears of AI manipulation as a ruse. He asserted the department is nervous and does not have a cogent legal stand to defend their decision.
However, Sen. Mark Warner, the Democratic chair of the Senate Intelligence Committee, recognized the threat of AI manipulation but suggested that the audio should be public with “watermarking components” or digital safeguards to detect resultant tampering. Notably, while the DOJ is careful not to release the audio, ironically, some Republicans have openly speculated that the released transcript does not accurately portray the interview.
AI disinformation and deepfakes have had a significant disruptive influence on global politics, with politicians often using the ease of creating fake data as a handy excuse when they find themselves in controversial situations. The issue of deepfakes poses a pressing concern for democracy and raises the question of how best to authenticate and secure public information.
