Republican candidate for Massachusetts Governor Brian Shortsleeve is facing backlash after a recent campaign ad used artificial intelligence to create a fake clip of his opponent, Democrat Maura Healey.
The ad, posted to social media, presents itself as a radio ad for Healey's re-election. In it, an AI-generated version of Healey's voice disparages her administration.
"We have one of the highest electricity rates in the nation," the AI Healey voice says with an almost imperceptible robotic tinge. "Thanks to me slapping on excessive fees to fund my climate agenda."
The ad does not disclose that it was made with AI.
Under the post on the Shortsleeve's Instagram page, a caption reads "here's what one of her radio ads might sound like - if she was honest."
In a statement, the Shortsleeve campaign said the ad was intended to be satire: "While this ad is a parody, there is nothing funny or fake about its substance, which highlights Maura Healey’s failed record of killing jobs and making Massachusetts the most expensive state in the nation."
The ad prompted pushback from the Massachusetts Democratic Party.
"SlowZone Shortsleeve has to create alternative realities to pretend he has any shot in this race," Chairman Steve Kerrigan said in a statement.
"He should stop lying to voters and tell them the truth—he will be a rubber stamp on President Trump's harmful agenda," Kerrigan said.
The campaign ad raises questions about the legal limits of AI technology in elections, especially so-called 'deepfakes': content made with artificial intelligence that feature fake digital versions of real people.
Deepfakes are regulated in political campaigns in 26 states, according to the National Conference of State Legislatures. Most of those laws only require AI-generated campaign ads to include a disclosure that they are not real, though some outright ban using the likeness of any candidate without express consent.
Massachusetts lawmakers have introduced a bill that would prohibit political deepfakes around election season, as a part of a broader effort to prevent AI technology from being used to spread election misinformation. But the bill only prohibits "materially deceptive election-related communication" that deliberately misleads voters on things like poll locations and political endorsements.
It does include a carve-out that would provide a path for a person whose likeness is used in a deepfake to sue for damages, but the bill does not apply to satire or parody.