Pennsylvania has imposed disclosure duties on certain synthetic-media advertisements, but the measure Governor Josh Shapiro signed on October 22 is narrower than some coverage implied. The Artificial Intelligence Deceptive Advertising Act, Act 99 of 2026, takes effect December 21, sixty days after signing. It targets deceptive commercial use of AI, not AI use itself.
- Key Point: The advertiser must act knowingly, recklessly or intentionally, and synthetic content must create a false perception about the nature, origin or authenticity of advertised goods or services. If AI does not mislead about the product, no warning is required.
- Key Point: Text-only ads are exempt under Section 4(b)(3), contradicting The Verge’s report that the law covers images, video, audio and text. Obviously fictional material, parody, satire, commentary and criticism are also excluded.
- Key Point: Civil fines cannot exceed $3,000, plus attorney fees and costs. Only the attorney general or the competent district attorney may enforce; consumers cannot sue.
- Key Point: Political advertising, platforms and agencies are absent: “political,” “candidate” and “election” never appear; social networks, hosting services and agencies go unmentioned. Only the advertiser is addressed, and it is undefined.
What a warning requires
Where the duty applies, the notice must state that the ad contains AI-modified or AI-generated content, appear when the content is first shown, use the same medium, and—if technically feasible—be permanent or extremely hard to remove. Visibility is measured by size, color, contrast, location, duration and audibility, in the consumer’s language, without contradicting the ad. Sponsor Nick Pisciottano, a Democratic state senator, said:
Consumers deserve to know whether AI is being used to create a false impression about a product or a service, and the law sets a common-sense transparency standard that lets people decide with information without stopping businesses from taking advantage of the technology.
Enforcement, lobbying and the election gap
With a $3,000 ceiling and no private right of action, the law is a weak deterrent against AI ad scams. Amendment A03233, expressly backed by the state’s business lobby, pulled the bill out of Pennsylvania’s broader consumer protection statute and made it standalone. The Pennsylvania Chamber of Business and Industry memo, signed by governmental affairs director Aaron Riggleman, welcomed the change:
The amendment narrows the scope of the law, clarifies the disclosure requirements and adds exemptions, and limits enforcement to the attorney general and district attorneys instead of opening the door to the expansive liability of the general consumer protection law.
It said more changes may be needed. Votes were 48 to 2 in the Senate in June, 163 to 40 in the House on October 7, and unanimous in all four committees. NetChoice, a large-platform trade group, wrote on October 4 that a parallel AI regime atop existing consumer laws does not improve protection and that prescriptive rules risk choking short-form video, small display ads and audio spots. MediaPost read the letter; the original is inaccessible. HB 811, which would ban fraudulent AI in election campaigns and create civil liability for misrepresenting a candidate, passed the House and has sat in the Senate Communications and Technology Committee since June 24, 2025. Pennsylvania thus regulates product ads, not candidate ads.
Where Pennsylvania sits
Pennsylvania is third by signature, second by effective date. Its Senate calls it “one of the first states”; ranking depends on criteria. New York was first: Governor Kathy Hochul signed its synthetic-performer law in December 2025, effective June 2026. It is narrower—only non-identifiable synthetic human figures, with audio ads excluded—but fines are $1,000 first and $5,000 later. California signed second through SB 1050 in September 2026, requiring a warning when a synthetic performer appears prominently; it is enforceable January 1, 2027, eleven days after Pennsylvania. Hawaii chose prohibition, banning realistic digital imitations of identifiable people without consent.
Act 99 is not a broad shield against AI advertising fraud. A bad-faith advertiser has little reason to disclose, and the modest penalty plus public-only enforcement leaves consumers with limited recourse. The law’s practical reach is confined to synthetic content that misleads about a product or service—and its exemptions and limits keep that reach narrow.