The Artificial Intelligence Deceptive Advertising Act, signed by Gov. Josh Shapiro on Oct. 22, mandates that advertisers disclose when AI-generated content in an ad creates a "false consumer perception." The law was sponsored by state Sen. Nick Pisciottano, a Democrat representing parts of Allegheny County in Senate District 45, who said his goal was to protect consumers as AI-generated images and videos become increasingly difficult to distinguish from reality.

"It was becoming more and more difficult to distinguish photos and videos that were real, created with traditional means, or created by AI," Pisciottano said. "And I felt that that was a danger to consumers, and it would make it easier for people to get scammed or ripped off."

Shapiro framed the law as a transparency measure. "Pennsylvanians deserve to know if and when AI is being used in ads," he wrote in a Facebook post announcing the signing.

But the law itself is narrower than the sponsor's public statements suggest. It applies only to AI-generated content that creates a "false consumer perception" — a phrase that is not defined in the statute. The "false perception" language was added in a later amendment; the bill was originally conceived as a mandate that all AI content, not just advertising, include a disclosure.

"What the law says and what its goal might be could be different things," said Libby O'Neill, an advertising and AI attorney with the law firm Loeb & Loeb. "The sponsor has described this as, 'If it's AI, label it.' The law, on the other hand, only requires a label if that AI use creates a false perception about the product, and we don't really know what 'false perception' means in this context."

Eleanor Huyett, an assistant professor at Temple University's Beasley School of Law who specializes in consumer protection, said the ambiguity makes compliance difficult to assess. She offered the example of an ad featuring an AI-generated hamburger that accurately reflects the product's ingredients and size but makes it look more appetizing than in real life.

"My honest answer is that we don't know yet because of that ambiguity," Huyett said. "If you're the advertiser, when in doubt, you want to err on the side of disclosure."

O'Neill agreed that caution is warranted. "Until this is tested in court, the conservative path is simple," she said. "If AI has, in the words of the statute, 'significantly modified' or generated the content in your ad, label it."

The law also leaves open who counts as an "advertiser" responsible for adding the disclosure. Pisciottano said he believes the responsibility falls on the technology companies that generate the content — for example, if a small business uses ChatGPT to create an ad, he believes ChatGPT, not the business, would be required to add the disclosure.

"The ultimate goal of this legislation is not to put that requirement on the business using the advertisement," Pisciottano said. "The onus of this is to require the tech companies themselves ... to basically self-disclose."

But O'Neill said the text does not support that reading. "As lawyers, we look at the text, and the law applies to 'advertisers' without defining the term," she said. "Read plainly, that would generally include the business running the ad. The senator's view may well shape how the law is enforced, but in the end the attorney general, the district attorneys and, ultimately, the courts will decide who's covered."

Enforcement will be carried out by the state attorney general or local district attorneys, with violators facing civil fines of up to $3,000 plus attorney fees and court costs. The law does not specify whether fines would be issued per advertisement, per violation or per instance of dissemination. Huyett said that approach is typical for consumer protection statutes, with agencies focusing resources on the most egregious offenders and patterns of complaints.

The law's passage comes amid growing public backlash over AI-generated images in local advertising. In Philadelphia, restaurants have faced online criticism after being accused of using AI images in signage, social media posts and menus. One Instagram influencer who reports on the city's food scene under a pseudonym and a Mexican wrestling mask said AI-generated restaurant ads undermine consumer trust.

"It becomes indistinguishable what's real [and] what's fake," the influencer said. "How can anyone decide they want to eat out when they're getting sold a bill of false goods even from the initial look of the place?"

Pisciottano acknowledged that the law's application will evolve as the technology does. "This is emerging technology and will continue to change," he said. "It's hard to predict the future contours of what the enforcement of AI-generated content will look like."

Huyett said such ambiguity is not unusual for laws regulating new technologies. "Often, when you're trying to draft a consumer protection statute like this one, you want your language to be broad enough that it can capture unforeseen behavior," she said. "Typically where this will get worked out is in the courts."

O'Neill noted that a similar New York law requiring disclosure for AI-generated "synthetic performers" in ads has been in effect since June without any enforcement cases, suggesting Pennsylvania's law may also take time to be tested. "We're going to have to see what kind of appetite there is on the enforcement side," she said.