Estimated reading time: 6 minutes · Last updated:
Pennsylvania now requires advertisers to label ads that were made or heavily altered with artificial intelligence. Governor Josh Shapiro signed the Artificial Intelligence Deceptive Advertising Act on Thursday. It covers still images, video, audio and text alike, and the disclosure must appear in the same format as the ad. The Pennsylvania AI advertising law applies no matter where an ad was produced, so a commercial made in another state but aired in Pennsylvania needs the label too. State Senator Nick Pisciottano introduced the bill. The measure reaches from national TV campaigns to a local restaurant's flyer. Yet its key terms are loosely drawn, including what counts as significantly modified, and it exempts satire and fiction.
Key takeaways
- Signed into law: Governor Josh Shapiro signed Pennsylvania's Artificial Intelligence Deceptive Advertising Act on Thursday. It makes disclosure mandatory for ads generated or significantly modified by AI.
- Every format covered: Pennsylvania's rule applies to still images, video, audio and text ads, and each label must take the same format as the ad it sits on.
- Origin does not matter: Under the act, an ad produced outside Pennsylvania still needs the label when it is intended to air in the state. That means the disclosures could surface in other states as well.
- Carve-outs: Parody, satire, commentary, criticism and evidently fictional content are exempt. Beyond those exemptions, the act applies only when synthetic content misleads consumers about a product or service.
- Bill sponsor: State Senator Nick Pisciottano introduced the Artificial Intelligence Deceptive Advertising Act. His office did not immediately answer questions about it.
Table of contents
What Pennsylvania's AI Ad Law Actually Requires
The statute is triggered by any advertising content that "has been significantly modified or generated by artificial intelligence." Once an ad meets that test, it must carry a disclosure. Pennsylvania did not pick out a single medium for the rule. A printed picture, a video spot, an audio clip and a block of written ad copy all fall inside its scope.
The disclosure also has to match the medium of the ad it accompanies. A label on an audio spot would therefore need to be heard, while one on a poster or flyer would need to be seen. Format-matching closes an obvious gap, because an AI-voiced radio ad with a text-only disclaimer somewhere online would tell a listener nothing.
Size and budget make no difference either. Terrence O'Brien, weekend editor at The Verge, noted that the rule reaches television commercials from companies worth billions just as it reaches a restaurant's crowded flyer promoting its wing specials. A local business owner who uses an image generator for a menu promotion faces the same disclosure duty as a national brand.
Why Ads Made Elsewhere Will Need Pennsylvania Labels
The requirement follows the ad into Pennsylvania instead of staying with the place where it was produced. Under the law, labels are needed regardless of where an ad was created. An agency working in another state on an AI-assisted spot for a campaign that includes Pennsylvania therefore falls under the rule.
That has consequences for audiences outside the state. O'Brien pointed out that the disclosures may appear in commercials shown in other states when those same ads are also expected to air in Pennsylvania. If an advertiser decides that one compliant version is simpler than separate cuts for each state, the Pennsylvania label travels everywhere the ad runs.
O'Brien also counted Pennsylvania among the earliest states to make AI-generated content carry a disclosure. Being early matters more when a law reaches beyond its borders. If other states do not set their own rules, a single state's labeling standard could end up shaping what national campaigns look like in practice.
Where the Exemptions Leave Room to Argue
The statute exempts "parody, satire, commentary or criticism," and it also excludes anything that is "evidently fictional." Neither phrase comes with a clear test. An AI-generated cartoon mascot might look plainly made-up to one viewer and like a believable product demonstration to another.
The threshold for coverage is just as soft. O'Brien called the meaning of "significantly modified" subjective. A retouched product photo, an audio track cleaned up with AI and a fully synthetic video sit at very different points on that scale. On his reading, the wording does not settle where the line falls.
A further qualifier narrows the law again. On O'Brien's reading, a disclosure would be triggered only by material that "uses synthetic content in a manner that creates a false consumer perception", and only when that misleading impression concerns a product or service. He flagged two questions this leaves open: how the law defines falsehood, and who gets to decide.
What Pennsylvania Has Not Yet Explained
The unresolved definitions land hardest on the businesses that must comply. An advertiser weighing whether a lightly edited image needs a label has three soft tests to apply at once: significant modification, evident fiction and false perception. A wrong call could mean a missing label where one was required, or an unneeded label that makes a campaign look less trustworthy.
Pisciottano, the state senator behind the measure, did not immediately respond to a request for comment. Until his office or the state's enforcers issue guidance, advertisers have only the statutory language to work from.
Several practical details also need checking against the bill itself: when the requirement takes effect, which agency enforces it and what a violation costs. Those answers will decide how quickly agencies, platforms and small businesses change the way they produce and approve ads.
| Provision | Statutory wording | Open question |
|---|---|---|
| Coverage trigger | Significantly modified or generated by artificial intelligence | How much AI editing counts as significant |
| Speech exemption | Parody, satire, commentary or criticism | Who decides whether an ad qualifies |
| Fiction exemption | Evidently fictional | Evident to whom |
| Deception qualifier | Creates a false consumer perception | What counts as false, and who rules on it |
Will the label requirement work?
The case for
- Format-matching means an audio ad gets an audible disclosure and a visual ad gets a visible one, so the label reaches the audience that actually encounters the ad.
- Because the rule covers ads regardless of where they were made, national advertisers cannot avoid it by producing campaigns out of state.
- Ads expected to air in Pennsylvania could carry labels in other states too, which spreads the disclosure beyond one jurisdiction.
The case against
- The meaning of significantly modified is subjective, which leaves room to argue that heavy AI editing falls below the threshold.
- Exemptions for parody, satire, commentary, criticism and evidently fictional content come without clear tests.
- Limiting coverage to content that creates a false consumer perception adds another judgment call, with no stated decision-maker.
What to be careful about
- Advertisers, including small local businesses, may misjudge whether an AI-edited ad crosses the significantly modified threshold.
- Broad exemptions for satire and evidently fictional content could be used to avoid labeling ads that viewers find misleading.
- Without published guidance from the sponsor's office or enforcers, compliance decisions rest on undefined statutory terms.
- Out-of-state agencies producing campaigns that air in Pennsylvania may not realise the rule applies to them.
The bottom line
Pennsylvania has written a simple principle into law: if AI significantly shaped an ad, say so, in the same medium the ad uses. The rule follows an ad into the state wherever it was made, so it reaches national campaigns, and viewers in other states may start seeing the labels too. The weakness lies in the definitions. Significantly modified, evidently fictional and false consumer perception each require a judgment call, and nobody has yet said who makes it. Until guidance or enforcement fills that gap, the law's real effect depends on how cautious advertisers choose to be. Careful brands will label generously, while others may test how far the exemptions stretch.
What to watch
- Watch for guidance from Senator Nick Pisciottano's office or Pennsylvania enforcers on what significantly modified and evidently fictional mean in practice; no date has been set.
- Watch for the first national commercials carrying Pennsylvania-compliant AI labels, which may appear in other states as well; no timeline has been given.
- Watch for the effective date and enforcement provisions in the bill text to be confirmed, since they determine when advertisers must start labeling.
Frequently asked questions
Does a small business flyer need an AI label in Pennsylvania?
It can. The Artificial Intelligence Deceptive Advertising Act covers still images and text, and O'Brien noted it reaches a restaurant's flyer about wing specials just as it reaches big-brand TV ads. A label is required if the content was generated or significantly modified by AI and creates a false consumer perception.
Is AI-generated satire exempt from Pennsylvania's ad disclosure rule?
Yes. The law carves out parody, satire, commentary or criticism, along with content that is evidently fictional. The statute does not spell out who decides whether a given ad qualifies for those exemptions.
Who introduced Pennsylvania's AI advertising bill?
State Senator Nick Pisciottano introduced the bill, and Governor Josh Shapiro signed it on a Thursday. Pisciottano's office did not immediately respond to a request for comment after the signing.
Related reading
