In October 2026 Pennsylvania became one of the first states to write AI-ad disclosure into law. Senate Bill 806 doesn't require labeling every AI ad — it requires a clear, conspicuous disclosure when AI-generated or AI-modified content creates a false impression about what's being sold. This guide covers what the law requires, its format rules and exemptions, who it's likely to reach, and how it stacks on top of the platform labels and FTC rules you already have to clear — then what a disclosure duty that keys on production actually changes about an AI ad-creative workflow.
In October 2026 Pennsylvania became one of the first US states to put AI-ad disclosure into statute rather than platform policy. Senate Bill 806 passed the Senate 48-2 on June 10, 2026 and the House 163-40 on October 7, 2026; Governor Josh Shapiro signed it on October 22, 2026, as Act 99 of 2026. By its terms it takes effect 60 days after signing — December 21, 2026. The rule is narrower than 'label every AI ad.' It applies when an advertisement uses AI-generated or AI-modified content in a way that creates a false impression about the nature, origin, or authenticity of the goods or services being sold, and in that case the ad needs a clear, conspicuous disclosure that the content was generated or altered with artificial intelligence.
It is not alone: California, New York, and other states are building their own versions, which means AI ad creative now has to satisfy a growing patchwork of state statutes on top of the platform labels and the FTC honesty rules it already had to clear. This guide covers what SB 806 actually says, where the lines are still fuzzy, how it fits the wider patchwork, and what a disclosure duty that keys on production changes about how you make ad creative. For the step-by-step compliance task, the companion tutorial is how to comply with state AI-ad disclosure laws.
The covered material is 'synthetic content' — an image, video, audio clip, or text that has been significantly modified or generated by artificial intelligence. The duty is a clear and conspicuous disclosure stating that the advertisement contains content generated or altered with AI, and it attaches when that synthetic content creates a false perception about the nature, origin, or authenticity of the goods or services being advertised. One carve-out matters for the definition itself: an ad whose AI-touched content is limited to text is exempt outright, so the disclosure duty is really about image, video, and audio.
Enforcement sits with the Pennsylvania Office of Attorney General and county district attorneys, and a court can impose a civil penalty of up to $3,000 per violation plus attorney fees and court costs. That per-violation structure is the part to internalize for anyone running volume: a test with many non-compliant variants is many exposures, not one.
This is the most important and most misread part of the law. Using AI to make an ad is not, by itself, what triggers the label. A fully AI-generated product photo that honestly shows the product may not require a disclosure; an AI-altered clip that makes a product look like something it is not does. SB 806 is a consumer-protection measure about misleading impressions, not an AI-usage registry — the question it asks is whether the synthetic content misrepresents what is being sold.
The catch is that the key terms are subjective. 'Creates a false perception' and 'substantially modified' are left for enforcers and courts to apply, and the statute does not spell out who decides what counts as false. For high-stakes creative — regulated categories, big spends, anything where the AI materially changes how the product reads — the honest operating posture is to assume a clear label is cheaper than litigating the line, and to get counsel involved rather than guess.
SB 806 does not just ask for a disclosure; it specifies how it has to appear. The disclaimer must be in the same medium as the ad. On-screen disclosures must stay up long enough to be read, and audio disclosures must be in a volume and cadence a consumer can actually hear. And where feasible, the disclosure should be 'permanent or extraordinarily difficult to remove from the synthetic content.'
That last phrase is doing real work. It points away from a detachable caption, a platform metadata panel, or an 'About this ad' screen and toward baking the disclosure into the asset itself — burned into the video frame, composited into the image, built into the creative rather than bolted onto one distribution surface. The mechanics of doing that cleanly across a batch of variants are in the companion how-to.
SB 806's enacted text doesn't contain a standalone clause spelling out its reach over out-of-state advertisers — it only names who enforces it: the Pennsylvania attorney general and county district attorneys. State deceptive-advertising enforcement generally keys on where the misleading conduct lands, not where it originated, which is why most legal commentary on the law treats a national campaign reaching Pennsylvania consumers as fair game. For a creator or agency running national paid social, the prudent read is that a buy able to serve a Pennsylvania audience should be treated as in scope, and disclosure handled at the creative level rather than by trying to geo-fence a labeled and an unlabeled cut. This is the one area of the law worth confirming with counsel before assuming an out-of-state campaign is outside it.
The law carves out content that is evidently fictional; parody, satire, commentary, or criticism; and content whose AI-touched material is limited to text — categories where a reasonable viewer would not be misled about a real product, or where the law's disclosure mechanics (an on-screen or audible label) wouldn't apply anyway. Pennsylvania is also separately considering legislation specifically aimed at AI-generated deepfakes in political campaign ads, distinct from SB 806, which is a consumer-sales statute rather than an election law. The carve-outs above are narrower than they sound, though: 'evidently fictional' is not a license to skip disclosure on a realistic synthetic testimonial, and the subjectivity of the core trigger means a loosely-claimed exemption is a risk, not a safe harbor.
Pennsylvania is one piece of a fast-growing map. California's SB 1050, signed in September 2026, requires advertisers to disclose AI-generated synthetic performers in ads — a different trigger and a different scope from Pennsylvania's deception test (see the California SB 1050 breakdown). New York has moved to regulate AI in advertising as well, and more states are drafting. There is no single federal AI-ad-disclosure statute, so a national advertiser now faces overlapping state rules that differ in what they cover, how they define the trigger, and what they penalize.
The practical consequence of a patchwork is that you cannot build one 'compliant' creative to a single national standard. You build to the strictest rule you are exposed to, or you accept the operational cost of maintaining different cuts for different markets — and for most creators the first option is simpler, because a clear, legible AI disclosure rarely hurts an honest ad the way a regulator action does.
There are now three layers governing an AI ad, and they are not substitutes. The platform layer is policy inside one ad system — Meta's 'AI info' label and self-certification for social and political creative (covered in AI-generated ads disclosure and UGC creatives) and Google's advertiser declaration and 'How this ad was made' panel (covered in Google Ads AI-generated content disclosure). The federal layer is the FTC's honesty rules — the ban on fake and AI-generated testimonials, plus substantiation and material-connection disclosure. The state layer is statutes like SB 806.
Clearing one does not clear the others. A perfectly labeled Meta ad can still violate a state deception statute; a state-compliant disclosure does not cure a fabricated endorsement under FTC rules. The durable move is to treat disclosure and honesty as one combined pre-flight check that satisfies the strictest applicable rule across all three layers — the self-labeling discipline laid out in AI content labeling applies directly here.
If you generate ad creative with AI at any volume, the real lesson of SB 806 is that compliance can no longer be a manual afterthought bolted on per creative in the ad platform. The law keys on two things you control at production time: what was AI-generated or modified, and whether a clear, hard-to-remove disclosure rides with the asset. Both are cheapest to handle in the pipeline, before an asset ever reaches an ad account.
Three habits make that work at scale. Keep a record of what was AI-generated or AI-modified, so you can answer 'is this in scope' without reconstructing it later. Decide disclosure at production and render the label into the asset where it is required, rather than relying on one platform's toggle. And keep a human review gate before publish, where the in-scope call, the label, and the honesty of every claim get checked once per creative. Volume is the risk multiplier — a fifty-variant test is fifty compliance decisions, and the account that makes those decisions ad-hoc is the one that ships an unlabeled variant into Pennsylvania.
The pipeline is the cheapest place to carry a disclosure duty, which is exactly where a content engine like Kompozy earns its keep — not as a labeler, but as the layer where AI creative is generated, kept on record, and sent out through one controllable gate. Kompozy is a generation and multi-platform publishing engine: it produces across 18 output formats — persona and avatar video, clips, carousels, images, text, blogs, newsletters — so the AI ad creative is made on-record as AI-made rather than scrubbed of its provenance to dodge detection.
Two design choices line up with what a deception-and-honesty regime wants. First, Kompozy generates from a declared AI Influencer persona pool governed by a written Persona Brief: the on-screen presenter is a declared brand identity with claim boundaries and banned-word rules, not a counterfeit stranger staged as a real customer — the honesty posture the FTC layer rewards. Second, it publishes the same asset across eight social platforms plus blog and email behind a per-post review gate, which is a natural single checkpoint to run the in-scope, label, and honesty check once per creative instead of per platform.
Be clear about the boundary. Kompozy produces and distributes on-brand creative and gives you the review gate; it does not run your paid ad account, make the legal call on whether a given creative creates a false impression, or file a platform's self-certification for you. Those judgments stay with you and, for high-stakes work, your counsel. For the hands-on version — deciding scope, writing a same-medium disclaimer, and compositing it so it survives — follow the how-to on complying with state AI-ad disclosure laws.
A clear and conspicuous disclosure on an advertisement when its AI-generated or AI-modified content creates a false impression about the nature, origin, or authenticity of the goods or services being sold. The label must state that the content was generated or altered with artificial intelligence. It is not a blanket 'label every AI ad' rule — the duty is tied to a misleading impression about the product.
Senate Bill 806 passed the Senate 48-2 on June 10, 2026 and the House 163-40 on October 7, 2026. Governor Josh Shapiro signed it on October 22, 2026, making it Act 99 of 2026. It takes effect 60 days after signing — December 21, 2026.
The statute itself doesn't spell out a territorial rule, but enforcement runs through the state attorney general and county district attorneys — the officials who act on deceptive advertising reaching Pennsylvania consumers, regardless of where an ad was produced. For a national paid-social campaign, the safer assumption is that reaching a Pennsylvania audience puts the creative in scope; confirm with counsel if a campaign is run only outside the state.
Only the ones where the synthetic content creates a false impression about the product. Using AI to make an honest ad is not, by itself, the trigger, and ads whose AI-touched content is limited to text are exempt outright. Because 'creates a false impression' and 'substantially modified' are otherwise subjective, the practical posture for most advertisers is that a clear label is cheaper than arguing the line.
Those are platform policies that apply inside one ad system; SB 806 is a state statute enforced by the attorney general and county district attorneys, with civil penalties up to $3,000 per violation. They stack rather than substitute. Clearing Meta's 'AI info' label or Google's 'How this ad was made' panel does not satisfy a state statute, and labeling media as AI does not cure a deceptive claim under FTC rules.
Pennsylvania's Senate Bill 806 requires a clear, conspicuous disclosure on advertising whose AI-generated or AI-modified content creates a false impression about the nature, origin, or authenticity of what's being sold. It covers image, video, and audio (text-only ads are exempt), was signed into law on October 22, 2026 as Act 99, and is enforced by the state attorney general with fines up to $3,000 per violation. It is one of several state laws, not a federal standard.
Get started → · ← All guides · Compare Kompozy vs other tools