Pennsylvania Passes Bill to Regulate AI Deepfakes in Elections

Pennsylvania’s House of Representatives has unanimously approved a bill aimed at regulating the use of artificial intelligence (AI) “deepfakes” in election campaigns. This legislation reflects a growing nationwide concern over the potential misuse of this technology, which can create hyperrealistic videos, images, and audio often designed to mislead the public. State Representative Tarik Khan, the bill’s primary sponsor, emphasized the need for updated regulations to protect voters as technological advancements continue to evolve. “It’s becoming so hard to tell what is real and what’s not real,” Khan stated.

The bill, known as HB 811, requires campaign advertisements featuring deepfakes to clearly disclose that the content is not genuine if disseminated within 90 days of an election. Violators could face civil penalties, with fines reaching up to $15,000 per day for municipal races and $250,000 per day for presidential or congressional contests. This legislation positions Pennsylvania among the more than two dozen states that have enacted laws to combat the misuse of deepfake technology in political contexts.

Growing Concerns Over Deepfake Influence

Deepfake advertisements have already appeared in various local, state, and national elections. A notable case involved a deepfake ad in New Jersey’s gubernatorial primary, while the former mayor of Shreveport, Louisiana claimed a deepfake depicting him as a failing student directly contributed to his electoral defeat in 2022. Another incident occurred in North Carolina, where a Democratic donor used AI to manipulate the voice of Republican candidate Mark Robinson, making him appear to endorse controversial statements.

Perhaps the most alarming example occurred with a deepfake robocall impersonating President Joe Biden, which urged New Hampshire voters to abstain from participating in the 2024 presidential primary. Although the individual responsible was acquitted of serious voter suppression charges, the telecom company involved faced a $1 million fine for broadcasting the message.

While previous instances of deepfakes have not decisively altered the outcomes of elections, the potential for future misuse looms large as AI technology becomes more sophisticated. Khan noted that the integrity of elections must be maintained as technological capabilities advance.

Legislative Progress and Public Confidence

The bill now moves to the state Senate for further consideration. Governor Josh Shapiro has already signed multiple laws addressing deepfakes, including one that classifies them as a form of digital forgery. This legislative momentum follows the U.S. Senate’s recent decision to remove a proposed 10-year moratorium on state regulation of AI from a federal spending bill, indicating a growing recognition of the need for regulatory frameworks.

Representative Jeremy Shaffer, a co-sponsor of the bill, highlighted the importance of focusing on election integrity, given the public’s fluctuating confidence in Pennsylvania’s electoral processes, especially following the controversies surrounding the 2020 presidential election and the events of January 6, 2021. Although there was no evidence of widespread fraud, accusations of a stolen election have persisted, influencing public perception and discourse.

While some legislators, including Rep. Paul Takac, believe the bill is a positive first step, they advocate for further enhancements. Takac expressed concerns about the potential for bad actors to exploit weaknesses in the electoral system. He suggested expanding regulations to encompass malicious deepfakes created by individuals unaffiliated with campaigns.

Public interest groups have also weighed in on the legislation. Mekela Panditharatne, senior counsel for The Brennan Center for Justice, described Pennsylvania’s bill as a necessary measure in the battle against AI misuse in elections. She noted that state legislation has likely deterred the proliferation of harmful deepfake ads thus far.

The rise of deepfake technology has generated a phenomenon known as “The Liar’s Dividend,” where the public’s awareness of deepfakes may lead to increased skepticism of all digital content, even legitimate information. This could have serious implications for public trust and voter education. Panditharatne stressed the need for balanced public education efforts to avoid fostering unwarranted distrust in authentic information.

Khan remains optimistic about the future of AI, advocating for continued innovation while emphasizing the necessity of protective measures. “We went from using the abacus to computers. It’s that big of a leap,” he concluded. As the bill awaits consideration in the Senate, the ongoing evolution of AI technology will likely continue to raise complex questions around regulation, public trust, and electoral integrity.