Pennsylvania Requires AI-Generated Ads to Carry Labels When Consumers Could Be Misled
Pennsylvania Governor Josh Shapiro has signed the Artificial Intelligence Deceptive Advertising Act into law, creating a labeling requirement for advertising that is generated or significantly modified by AI. Shapiro announced the signing in a post on his official Facebook page dated October 23, 2026, describing the measure as a bipartisan bill that requires transparency.
The Act requires any ad in still image, video, audio, or text format that has been significantly modified or generated by artificial intelligence to carry a label. The label must appear in the same format as the ad itself, so a video ad carries a disclosure in video, and a text ad carries one in text. The Verge reports that the disclosure requirement applies regardless of where an ad was created, which means the state's rule is not tied to the geography of production.
The Act carves out several categories. Parody, satire, commentary or criticism, and anything that is evidently fictional fall outside its scope. Only synthetic content used in a manner that creates a false consumer perception about a product or service would require a disclosure. That qualifier sets the operative test: a disclosure is triggered by the likelihood of deception about what is being sold, not by the mere use of generative tools, meaning models that produce new images, audio, or text, somewhere in the production pipeline.
Pennsylvania State Senator Nick Pisciottano introduced the bill. Senate Bill 806 establishes the Act, defines it as a matter of unfair methods of competition, and provides for enforcement and penalties, according to the General Assembly's legislative records. The Act takes effect 60 days after it was signed into law, according to CBS Pittsburgh, which also describes Pennsylvania as one of the first states to require disclosures on AI-generated ads.
A separate measure, House Bill 95, would prohibit a person from knowingly or willfully disseminating an advertisement containing synthetic advertising. That bill is a proposal in the legislative record and does not share the Act's enacted status. The state's approach to synthetic media in advertising is still taking shape across both chambers.
The broader context here is familiar to anyone who has worked on consumer protection in digital media. Labeling regimes depend on definitions, and the Act's central terms, "significantly modified" and "false consumer perception," will carry most of the practical weight. Compliance teams at agencies, ad platforms, and brands will need to decide how much editing or generation crosses the threshold, and enforcement under the unfair competition framework will ultimately define where that line sits.
In my view, the most useful feature of the text is that it ties disclosure to consumer harm rather than to the tool used. Plainly satirical or fictional work remains untouched, and the obligation focuses on cases where a viewer or reader could reasonably be misled about a product. For engineers building generation and editing pipelines, that distinction matters more than the headline. Provenance metadata, the embedded record of how a file was made, and format-native labels are the kind of infrastructure that can make compliance routine, and a clear standard gives vendors a concrete target to build against.
