Pennsylvania has become one of the first U.S. states to require disclosure labels on AI-generated advertising. Governor Josh Shapiro signed the Artificial Intelligence Deceptive Advertising Act into law, establishing a new baseline for transparency in commercial content. The law is real, the intent is clear, and the loopholes are significant. Understanding all three matters more than the headline.
What the Law Actually Requires
The legislation covers a broad range of advertising formats: still images, video, audio, and text. Any ad that has been “significantly modified or generated by artificial intelligence” must carry a disclosure label, and that label must appear in the same format as the ad itself. A TV commercial requires an on-screen label. An audio ad requires an audible one. A printed flyer requires a printed notice.
The scope is deliberately wide. It applies to large corporations running national campaigns and to small businesses distributing local promotional materials. It also applies regardless of where the ad was originally produced. If a commercial is expected to air in Pennsylvania, the disclosure requirement follows it, even if the ad was created elsewhere. This means audiences in other states may begin seeing AI disclosure labels on content produced for multi-state distribution, simply because Pennsylvania is part of the broadcast footprint.
That breadth is one of the law’s genuine strengths. It does not create a carve-out for large advertisers or for content produced outside state borders.
Where the Law Gets Complicated
Here is what most coverage of this law underplays: the definitions that make or break enforcement are left largely undefined.
The central trigger for disclosure is whether content has been “significantly modified” by AI. That phrase does the heaviest lifting in the entire law, and it is entirely subjective. A minor color correction applied through an AI tool is different from a fully synthetic product image, but the law does not draw that line. Who decides what counts as “significant” is not specified.
The law also carves out exceptions for parody, satire, commentary, and criticism, as well as content that is “evidently fictional.” These are reasonable categories in principle. Satire has always occupied a protected space in advertising and media. The problem is that “evidently fictional” is not defined. Fictional to whom? Under what standard? A surrealist ad for a real product could be read as evidently fictional by its creators and as misleading by a consumer.
There is a further complication in the law’s framing around consumer harm. The wording suggests that disclosure is required only when synthetic content “creates a false consumer perception” about a product or service. This introduces a harm threshold: not all AI-generated content triggers the rule, only content that actively misleads. That is a meaningful distinction, but it raises an immediate question. Who determines whether a false perception has been created, and at what point in the process? Before the ad runs, or after a complaint is filed?
State Senator Nick Pisciottano introduced the bill. His office did not respond to requests for comment on these questions.
Why This Moment Matters Beyond Pennsylvania
Legislation like this is significant not because it solves the problem of AI-generated advertising, but because it marks the point at which governments begin treating AI content as a category that requires its own rules. That shift has consequences that extend well beyond one state.
For advertisers and agencies, the law introduces a compliance layer that did not exist before. Teams working on campaigns distributed across multiple states now need to track not just content standards but disclosure requirements that vary by jurisdiction. The practical burden falls on the people who produce and approve advertising, and it requires them to make judgment calls about what qualifies as “significantly modified,” in the absence of clear regulatory guidance.
For consumers, the law represents an acknowledgment that AI-generated content in advertising is now common enough to warrant a public notice system. That is a meaningful signal. It reflects a broader recognition that the line between a photograph of a real product and a synthetic rendering of one is no longer obvious to the naked eye, and that people have a reasonable interest in knowing which they are looking at.
The deeper issue the law surfaces is definitional. Regulators are being asked to govern a technology that does not have stable, agreed-upon boundaries. “AI-generated” is not a binary state. Content exists on a spectrum from entirely human-made to entirely synthetic, with countless hybrid points in between. Any law that tries to draw a line on that spectrum will face the same challenge Pennsylvania’s law faces: the line is arbitrary until it is tested, and it will be tested by the cases that fall closest to it.
In Short
Pennsylvania’s AI Deceptive Advertising Act establishes a real disclosure requirement for AI-generated ads, covering all formats and applying regardless of where content is produced. Its reach is broad. Its definitions are not. The terms “significantly modified,” “evidently fictional,” and “false consumer perception” are all left open to interpretation, which means the law’s practical effect will depend heavily on how those terms are eventually applied and by whom. This is what early AI regulation looks like: a genuine attempt to set a standard, followed by the harder work of figuring out what the standard actually means.
Based on reporting from The Verge.