Consultant Fined $6 Million for Using AI to Imitate Biden’s Voice in Robocalls, According to Reuters

By David Shepardson

WASHINGTON – The Federal Communications Commission (FCC) has imposed a $6 million fine on a political consultant for creating deceptive robocalls that mimicked the voice of President Joe Biden, instructing voters in New Hampshire not to participate in the Democratic primary.

Steven Kramer, a Louisiana-based Democratic political consultant, faced indictment in New Hampshire in May for the calls, which featured an audio clip resembling Biden urging residents to refrain from voting until November. Kramer previously worked with Biden’s primary opponent, Representative Dean Phillips, who condemned the robocalls.

In January, Kramer informed media outlets that he spent $500 to send these calls to voters to raise awareness about the potential dangers of artificial intelligence in political campaigns.

The FCC indicated that the robocalls were produced using an AI-generated deepfake audio designed to sound like Biden. According to FCC regulations, misleading caller ID information is prohibited. The commission stated that Kramer must pay the fine within 30 days, or the case will be forwarded to the Justice Department for collection.

Attempts to reach Kramer or a spokesperson for comment were unsuccessful.

FCC Chair Jessica Rosenworcel remarked on the situation, noting the ease and low cost of using artificial intelligence to imitate voices, contributing to the rise of misleading sounds and images. She emphasized the significance of addressing these issues, as such technology could unlawfully interfere with elections.

In August, Lingo Telecom agreed to a $1 million settlement after the FCC found that it transmitted the deceptive robocalls in New Hampshire. Under this agreement, Lingo will implement a compliance plan to ensure strict adherence to FCC caller ID authentication guidelines.

Moreover, the FCC in July proposed a requirement for broadcast radio and television political advertisements to disclose whether the content was generated using AI, although this proposal remains under consideration.