After last month's primary election saw a message designed to sound like US President Joe Biden's voice telling New Hampshire residents not to vote, the Federal Communications Commission (FCC) is moving to make it illegal to use artificial intelligence-generated voices for robocalls. The agency is concerned that audio deepfake technology is being used to confuse and deceive consumers by imitating the voices of "celebrities, political candidates, and close relatives."
FCC Chairwoman Jessica Rosenworcel's proposal would make it illegal under the Telephone Consumer Protection Act (TCPA) to use artificial intelligence-generated voices to make pickup calls.
The 1991 TCPA law restricts telemarketing, the use of automatic telephone dialing systems and artificial or prerecorded voice messages. The law also requires telemarketers to obtain consumers' express written consent before making voice calls to consumers. The FCC wants AI-generated sounds to adhere to those standards as well. The five members of the committee are expected to vote on the proposal in the coming weeks.
"AI-generated voice clones and images are already sowing the seeds of confusion, tricking consumers into thinking scams and fraud are legitimate," Rosenworcel said in a statement. "No matter which celebrity or politician you like or what your relationship is with your relatives, we are all likely to be targeted by these fake calls when they call for help."
Speech fraud detection company Pindrop analyzed the 39-second fake message that sounded like Biden telling New Hampshire residents not to vote and found that it was created using ElevenLabs' text-to-speech engine. The startup uses artificial intelligence software to replicate the sounds of more than two dozen languages.
According to Bloomberg, ElevenLabs confirmed that its software was used in the New Hampshire incident and banned the account responsible. The New Hampshire Attorney General's Office has announced an investigation into phone calls from people impersonating Biden.
An FCC spokesman said potential changes to the law would give state attorneys general new powers to crack down on voice call scammers using artificial intelligence while protecting consumers.
The FCC has been fighting robocalls for years. In December 2022, the Federal Communications Commission proposed a $300 million fine against an illegal multinational robocall operation that placed more than 5 billion robocalls to more than 500 million phone numbers in a three-month period in 2021. In 2020, the bureau also imposed $225 million in fines for misleading robocalls.
Generating the voices of family members has become a common means for criminals to steal money from unsuspecting victims. This crime continues to increase as AI-generated technology becomes more accessible, cheaper, and more convincing.