Pennsylvania’s AI Ad Disclosure Law: What Marketers Need to Do Before December 21

Pennsylvania has signed a law requiring disclosures on certain AI-generated or altered advertisements. Governor Josh Shapiro approved Senate Bill 806 on October 22; its 60-day effective period ends December 21, 2026. This is not a blanket rule for every ad that used AI: it targets synthetic content used in a way that creates a false consumer perception about a product or service, leaving marketers judgment calls and little time to build a review process. Pennsylvania’s legislative record confirms enactment as Act 99 of 2026, while the bill text sets out the disclosure rule.

What Pennsylvania’s AI ad law actually covers

The law defines synthetic content to include images, video, audio, or text that has been significantly modified or generated by AI. Its disclosure requirement applies when an advertiser uses that content in a manner that creates a false consumer perception about the nature, origin, or authenticity of the advertised goods or services. That is narrower than “label every AI-assisted asset,” but broader than a rule limited to fake celebrity endorsements or deepfakes.

For example, a digitally altered product image that makes an item appear materially different from what buyers receive, or synthetic audio that implies a real person endorsed a service, could warrant review under the law’s standard. Those are risk scenarios, not automatic legal conclusions: the statute does not precisely define what makes a consumer perception “false.” A legal analysis from Loeb & Loeb highlights the interpretive uncertainty and cautions that a disclosure does not excuse a misleading underlying claim.

The statute expressly excludes content that is evidently fictional, parody, satire, commentary, criticism, or limited to text. For covered content, the disclosure must say that the advertisement contains material modified or generated using AI. It must appear the first time the synthetic content is presented, use the same medium, and, when technically feasible, be permanent or extraordinarily difficult to remove. The law authorizes the Pennsylvania Attorney General or an appropriate district attorney to seek an injunction or civil action; a court may award a fine of up to $3,000, plus attorney fees and costs.

Why this creates a workflow problem, not just a labeling task

AI is embedded in routine production: background replacement, product-scene generation, image-to-video, voiceover, localization, and edits mixing real product photos with generated elements. The question is not simply whether a team used AI, but whether synthetic content could mislead people about what the business sells, where it comes from, or whether a real person or event is involved.

That makes asset provenance essential. If an agency receives a finished ad without knowing what was altered, it may not be able to assess disclosure. Resizing or reformatting across video, audio, social, and display can also strip labels. The same-medium and first-presentation rules make version control part of compliance.

Geographic scope requires counsel’s attention. The law does not spell out a territorial test for campaigns built elsewhere and delivered into Pennsylvania. Businesses should not assume that being headquartered elsewhere resolves the issue; ask counsel how it applies to planned media and audiences.

A practical readiness plan before December 21

Marketing leaders do not need to halt AI creative. They do need a fast, documented decision path. Start with a focused inventory of current campaign assets that use generated or significantly altered images, video, audio, or product depictions. Prioritize ads that show product appearance, performance, origin, endorsements, or customer outcomes, because those are the claims most likely to affect a buyer’s understanding of what is being advertised.

Next, add four fields to the creative handoff: whether AI was used; what elements were generated or changed; which markets and placements will receive the asset; and whether review found a potential false-impression risk. Route uncertain cases to legal review rather than asking a designer or media buyer to interpret the statute alone. If counsel determines that a disclosure is appropriate, make it clear and conspicuous in the asset itself, in the same medium, at the first presentation, and preserve a record of the approved version. Do not assume a caption, landing-page note, or platform disclosure automatically meets those requirements.

Finally, assign one owner to check that the approved disclosure survives cropping, translation, placement changes, and agency-to-platform handoffs. Keep the decision and final creative with the campaign record. That gives the business a repeatable process if the same asset is reused or a regulator asks how the team assessed it.

The signal for national advertisers

Pennsylvania’s law is a concrete reminder that AI-generated marketing is moving from voluntary transparency principles toward enforceable ad rules. Its narrow trigger, broad creative formats, and ambiguous “false consumer perception” test make it a poor candidate for a one-line global policy. National brands should build a reusable review standard that can accommodate state-specific requirements without labeling every AI-assisted draft indiscriminately.

Real Internet Sales helps businesses turn emerging AI and digital-marketing changes into practical strategy, content, and campaign controls. Call 803-708-5514 or visit realinternetsales.com to discuss a responsible marketing plan for the AI era.