Pennsylvania just turned every marketer in the state into an AI auditor, whether they know it or not.
The Summary
- Pennsylvania Governor Josh Shapiro signed the AI Deceptive Advertising Act, requiring labels on any ad "significantly modified or generated by AI" — photos, video, audio, text
- Applies to every advertiser in the state, from Fortune 500 TV spots to local pizza shop flyers
- Pennsylvania becomes one of the first states to mandate AI disclosure in commercial speech, regardless of where the ad originates
The Signal
The law doesn't define "significantly modified." That's the whole game. Did you use an AI upscaler on a product photo? Label it. Did Canva's AI suggest a layout and you clicked yes? Probably label it. Did you touch up a background with generative fill? Definitely label it. Every creative decision now comes with a compliance question.
This isn't about deep fakes. The law treats a restaurant's AI-generated coupon the same as a synthetic video of a CEO. No threshold for harm. No carve-out for obvious uses like stock photos or basic editing. If AI touched it meaningfully, disclose it. The vagueness is the point — it forces advertisers to over-label rather than risk being wrong.
"Pennsylvania just made AI transparency the default cost of doing business."
The enforcement mechanism matters more than the text. Who verifies compliance? How do regulators prove an image was AI-modified versus human-edited with AI-assisted tools? Adobe's Firefly credits appear in metadata, but Midjourney outputs don't carry watermarks. Small businesses using free AI tools won't know they're non-compliant until they get a notice. Large advertisers will hire compliance teams. The gap between them just widened.
What Pennsylvania really did:
- Created a new category of commercial speech regulation that assumes AI is inherently deceptive
- Made every ad buy in the state carry legal risk for out-of-state companies that don't track Pennsylvania separately
- Set a precedent for other states to copy, creating a patchwork of conflicting disclosure rules
Pennsylvania's move follows broader state-level AI regulation efforts, but this one cuts across every industry. It's not about election integrity or consumer finance. It's commercial speech, which means it affects every brand that runs ads in Pennsylvania media markets or targets Pennsylvania consumers online. The compliance cost isn't the label itself. It's tracking which ads need them.
The timing is deliberate. Pennsylvania is a swing state that just lived through an election cycle where AI-generated political content was everywhere. This law doesn't touch political speech — that's protected differently — but it captures everything else. Campaigns can synthetic-video their way through Pennsylvania. Car dealerships cannot.
The Implication
If you're running ads in Pennsylvania, audit your creative workflow now. If AI touched any part of production, assume you need a label until a lawyer tells you otherwise. If you're a platform, start building state-specific compliance features into ad managers. If you're watching from another state, Pennsylvania just became the test case for whether AI disclosure laws scale or collapse under their own vagueness.
The real shift is cultural. Pennsylvania normalized the idea that AI-generated content is suspect by default. That's a harder problem to solve than compliance.