The White House is engaged in “active efforts” to better identify deepfakes and is urging researchers and companies to do more to combat manipulated media, a Biden administration official said Tuesday.
Anne Neuberger, deputy national security advisor for cyber and emerging technology, said during a Washington Post Live event that the White House held an event last week about “building defenses to counter AI-driven voice cloning,” and is exploring the possibility of watermarking computer-generated content.
“Companies are looking at and need to do more to build the technology to identify what are deepfakes,” Neuberger said. Watermarking may be one such method, and Neuberger urged “responsible” companies to explore the use of tools that would inform consumers what material is AI-generated.
Neuberger said she is looking to Congress, which is currently considering a raft of different proposals regarding AI governance to write the “new laws we need for this very new space” of manipulated media.
A White House official said in an email to CyberScoop that the administration recently “convened key stakeholders to discuss the state of AI-generated voice cloning technology and how we can work together to develop viable solutions to combat AI-enabled robocalls.” The discussion was a closed event.
Neuberger said the use of watermarking could be “particularly significant” for a White House that is increasingly “worried” about deepfakes, referencing a fake image of an explosion at the Pentagon that circulated among verified accounts on Twitter. Just last month, fake audio of President Joe Biden urging Democratic voters to not vote in the New Hampshire primary showed the ease with which AI can be used to impersonate public officials and potentially be used to influence elections in 2024.