Short answer
Pennsylvania has enacted a law requiring ads that are generated or significantly modified by artificial intelligence to carry a label in the same format as the ad itself. The rule does not depend on where the ad was made. Several terms remain vague, so brands advertising to US audiences should organise their creative records now.
Highlights
- The law covers still images, video, audio and text ads that AI generated or significantly modified.
- The label must use the same format as the ad, wherever the ad was created.
- Terms such as "significantly modified" and "evidently fictional" are left vague.

2 min readAuthor: UNIT Journal EditorEditor-in-chief: Uğur Deniz İlhan
As The Verge reports, Pennsylvania has signed the Artificial Intelligence Deceptive Advertising Act into law. It asks advertisers to disclose when an ad was generated by AI.
Which ads does the law cover?
Any ad that has been "significantly modified or generated by artificial intelligence" is in scope, whether it is a still image, video, audio or text. The label has to appear in the same format as the ad itself, so an audio ad needs an audio disclosure. The reach runs from television commercials by large companies to a local restaurant's flyer.
Where an ad was created does not matter. An ad expected to run in Pennsylvania may need a label even if it was produced in another state or another country.
What loopholes does the law leave?
The Verge points to several vague spots. What counts as "significantly modified" is subjective. Parody, satire, commentary, criticism and anything "evidently fictional" are carved out, but the last term is undefined. The wording also suggests that only synthetic content creating a false consumer perception of a product or service needs a disclosure, and it is unclear what counts as "false" or who decides.
What should brands advertising to US audiences do?
For brands and agencies that advertise to US audiences or prepare campaigns that could run in Pennsylvania, the first step is to record which creative assets involve AI. Building a cautious internal process until the limits become clearer is cheaper than pulling ads later.
- Record, for every asset (image, video, audio, text), whether AI was used and how, in your creative production log.
- Flag work that could count as significantly modified, such as AI-generated people, product visuals or voices, and prepare the label wording in advance.
- Plan the label in the ad's own format: on-screen text for video, a spoken notice for audio, a disclosure line for text ads.
- If Pennsylvania is in the target area, ask legal counsel beforehand whether the "evidently fictional" exception can be relied on.
If campaigns run through a performance marketing team, carrying the creative source record into ad-account naming and the measurement sheet also makes it easier to show later which version of an ad ran with a label.
Frequently asked
- Does the law apply to ads made outside Pennsylvania?
- According to The Verge, the label requirement applies regardless of where an ad was created, so ads expected to appear in Pennsylvania may need one. Legal counsel should confirm the position for a specific campaign.
Sources
- The Verge ·
Follow UNIT Journal
What's new in search, AI and technology, in your feed every day.


