Pennsylvania just became the first state to require disclosure labels on AI-generated advertising. Governor Josh Shapiro signed the Artificial Intelligence Deceptive Advertising Act on Thursday, and it’s broader than you might think.

The law covers any ad that “has been significantly modified or generated by artificial intelligence.” That means still images, video, audio, and text. It applies to a Fortune 500 company’s TV spot and to the corner diner’s Instagram post. If AI made it or substantially changed it, you have to say so.

The disclosure has to appear in the same format as the ad itself. A video ad needs a visual or audio disclosure. A print flyer needs text. The idea is that consumers shouldn’t have to hunt for the label or know where to look.

What counts as “significantly modified”

The statute doesn’t define “significantly modified” with precision, which is going to matter a lot in practice. Is it an AI-generated background? A face swap? Text written by ChatGPT? The law leaves room for interpretation, and that means enforcement will have to fill in the gaps.

Pennsylvania’s Attorney General can bring civil actions for violations. There’s no private right of action, so this isn’t going to spawn a wave of class actions. But the AG has broad authority to go after deceptive practices, and AI-generated ads that mislead consumers are squarely in that wheelhouse.

The penalties aren’t specified in the coverage of the law, but Pennsylvania’s consumer protection statutes typically allow for civil penalties and injunctive relief. The real teeth here aren’t necessarily in fines. It’s in the reputational cost of getting caught running undisclosed AI ads and the hassle of having to pull a campaign mid-flight.

Why states are moving first

This is a state-level response to a federal vacuum. Congress has held hearings on AI and advertising, but no federal disclosure requirement exists. The FTC has general authority over deceptive advertising, but it hasn’t issued specific rules about AI-generated content. That leaves states to act on their own.

Pennsylvania’s law is narrow. It doesn’t regulate AI use generally or impose restrictions on what you can say in an ad. It just requires disclosure. That makes it more likely to survive a First Amendment challenge, though someone will probably try. Compelled commercial speech has to serve a substantial government interest and be narrowly tailored. Preventing consumer deception clears that bar pretty easily.

Other states are watching. California has considered similar bills. New York has floated disclosure requirements for political ads specifically. If Pennsylvania’s law works without creating chaos, expect more states to follow.

Who this actually affects

Big advertisers already have compliance teams and can add a disclosure line to their creative briefs. The bigger impact is on small businesses and individuals who don’t think of themselves as advertisers but technically are.

If you run a small business and use an AI tool to generate your Facebook ads, you’re covered by this law. If you’re a real estate agent using AI to touch up listing photos or write property descriptions, same thing. The law doesn’t carve out exceptions for size or sophistication.

That’s going to create some friction. A lot of people using AI tools don’t realize they’re using AI, or don’t think it matters. Adobe’s generative fill, Canva’s magic write, even your phone’s photo enhancement features could all trigger the disclosure requirement depending on how they’re applied.

The law takes effect immediately, which means Pennsylvania businesses need to audit their ad practices now. If you’re using AI tools in your marketing, you need a process for adding disclosures. If you’re not sure whether something counts as “significantly modified,” the safe move is to disclose anyway.

What happens next

Enforcement is going to be the real test. The AG’s office will have to decide what counts as a violation worth pursuing and what’s too minor to bother with. Early enforcement actions will set the tone and clarify the boundaries.

There’s also a question of how consumers will react. Disclosure requirements assume people care about the information and will use it. But if every ad has a disclosure label, does it just become wallpaper? The effectiveness of this law depends on whether the labels actually change behavior or just become one more thing people ignore.

Pennsylvania made a bet that transparency is better than prohibition. The law doesn’t ban AI in advertising or restrict what you can generate. It just says you have to tell people. That’s a reasonable approach, and it’s probably the right one for a technology that’s evolving faster than regulators can keep up.

But reasonable doesn’t mean easy to implement. Small businesses, content creators, and anyone running ads in Pennsylvania now has homework to do. Figure out what you’re using, whether it’s covered, and how to disclose it. The law is in effect, and “I didn’t know” isn’t going to be a defense.