Pennsylvania signs law requiring labels on misleading AI-generated ads

Pennsylvania Governor Josh Shapiro signed Senate Bill 806, the Artificial Intelligence Deceptive Advertising Act, on Thursday, October 22. The law requires a disclosure on advertising content that "has been significantly modified or generated by artificial intelligence" when that content could mislead consumers about what is being sold. According to an analysis by the law firm Loeb & Loeb, it takes effect 60 days after signing, on December 21, 2026.

The Verge reports that Pennsylvania is one of the first states to require disclosures on AI content. The rule depends on where an ad is shown, not where it was made. A national commercial that will air in Pennsylvania may need a label everywhere it runs, unless the advertiser produces a separate version for the state.

What the law requires

The bill was introduced by State Senator Nick Pisciottano. According to Loeb & Loeb's reading of the enacted text, the disclosure must state that the ad "contains content that was modified or generated using artificial intelligence." It must be clear and conspicuous. It must appear the first time the synthetic content is shown, in the same medium as that content: a spoken label for audio, a visual one for images and video. Where technically feasible, it must be permanent or extremely hard to remove.

The obligation is not triggered by every use of AI. It applies when synthetic content is used in a way that "creates a false consumer perception" about the nature, origin or authenticity of the goods or services. The Verge reports that the law covers still images, video, audio and text. Loeb & Loeb's analysis says text-only content is exempt. Advertisers who rely on AI-written copy will want to check the final text on that point.

Two other categories are exempt: content that is "evidently fictional", and parody, satire, commentary or criticism.

Enforcement and penalties

According to Loeb & Loeb, only the state attorney general and district attorneys can enforce the law. They can seek an injunction or bring a civil action. A court may impose a fine of up to $3,000, plus attorney fees and costs. Consumers cannot sue under the law themselves.

The firm notes that the enacted text has no knowledge or intent requirement. Language limiting liability to violations committed "knowingly, recklessly or willfully" was removed during the Senate process. An advertiser can therefore be liable without intending to deceive. A label is also no defense in itself. If an ad misrepresents a product, other remedies still apply, including Pennsylvania's false advertising law and Section 5 of the FTC Act.

Pisciottano told CBS Pittsburgh that the law is meant to protect consumers. He pointed to reports from consumer groups of older residents being scammed through fake AI advertising. He described the disclosure as a line at the bottom of the ad, and said he expects federal action or other states to follow with similar rules.

The open questions

The Verge points to several loopholes. What counts as "significantly modified" is subjective. So is "evidently fictional". The "false consumer perception" test leaves open what qualifies as false and who decides.

Loeb & Loeb raises the same issue. According to the firm, a senator asked during debate "false as compared to what," and neither chamber changed the wording. The firm also says the enacted text is narrower than the broad disclosure rule Pisciottano has described in public. Under the text, an AI-generated presenter who gives accurate information may not need a label. An AI-altered real person giving a fake endorsement probably would. The attorney general and 67 district attorneys can each enforce the law, and there is no guidance or enforcement record yet. The firm expects advertisers to add labels whenever they are unsure.

The firm also compares the law with California and New York. Those states focus on synthetic performers. Pennsylvania's law covers any significantly AI-altered media and has no express exemption for expressive works or translations. Like New York, and unlike California, it has no private right of action.

The Verge says Pisciottano's office did not immediately respond to a request for comment.

What it means for work

The law adds a review step to advertising production. Someone has to record which images, voices and video were generated or significantly altered with AI, decide whether that use could create a false impression of the product, and place a label in the right medium on first appearance. For large brands and agencies, this means documenting AI use in creative assets, adding it to legal review, and possibly cutting a separate version for ads that air in Pennsylvania. For small businesses that make their own flyers and social posts with AI tools, the duty falls on the owner, often with no legal staff. Because the law has no intent requirement, it is not enough to say the tool did it. The biggest open question is where the line falls. Until enforcers or courts define "significantly modified" and "false consumer perception", the safest practice is probably to label more ads than the law strictly requires.