# Pennsylvania will make advertisers label AI ads that could fool shoppers > Source: > Published: 2026-10-24 19:33:00+00:00 Josh Shapiro, now Pennsylvania’s governor, speaking at a news conference in May 2022, when he was the state’s attorney general. File photo. Image: [Governor Tom Wolf (Flickr)](https://commons.wikimedia.org/wiki/File:Josh_Shapiro_2022.jpg) / Wikimedia Commons, [CC BY 2.0](https://creativecommons.org/licenses/by/2.0/), cropped Advertisers that use AI to make a product look better than it is will soon have to say so in Pennsylvania. Governor Josh Shapiro [signed the Artificial Intelligence Deceptive Advertising Act](https://www.palegis.us/legislation/bills/2025/sb806) on Thursday, October 22, putting the state among the first in the US to require labels on misleading AI-generated ads. The law, Senate Bill 806 (now Act 99 of 2026), was written by state Senator Nick Pisciottano, a Democrat from Allegheny County, with Republican Senator Tracy Pennycuick among its co-sponsors. It passed the Senate 48 to 2 in June and the House 163 to 40 on October 7. It takes effect 60 days after signing, around December 21. ## What advertisers have to label The [final text](https://www.palegis.us/legislation/bills/text/PDF/2025/0/SB0806/PN1808) covers ‘synthetic content’, which it defines as an image, video, audio clip or text ‘that has been significantly modified or generated by artificial intelligence’. An advertiser may not ‘knowingly, recklessly or willfully’ put out an ad that uses such content in a way that ‘creates a false consumer perception about the nature, origin or authenticity’ of what is being sold, unless it carries a clear and conspicuous disclosure. That label must say the ad contains content modified or generated using AI. It has to appear the first time the consumer sees the synthetic content, in the same medium (spoken in an audio ad, on screen in a video), and in the language the consumer is using. Where technically feasible it must be permanent or ‘extraordinarily difficult to remove’, a high bar given that [OpenAI and Meta have told US judges](https://madrobot.blog/2026/10/16/openai-meta-tell-us-judges-deepfakes-cant-be-reliably-detected-rule-901c/) there may never be a reliable way to spot AI fakes. Pisciottano’s office gave an example: an AI-generated image that could lead shoppers to believe they are seeing the actual product would need a label. ## ‘Consumers deserve to know’ Pisciottano pitched the bill as consumer protection when it cleared the legislature, saying in a [statement](https://pasenate.com/pisciottanos-landmark-ai-bill-establishes-one-of-nations-first-advertising-disclosure-standards/): Consumers deserve to know if AI is being used to create a false impression about a product or service. Nick Pisciottano, Pennsylvania state senator He said it sets ‘a commonsense transparency standard’ while ‘still allowing businesses to take advantage of this technology’, and that he was proud lawmakers had come together across party lines ‘to set a standard that other states can follow’. ## The loopholes The law is narrower than its name suggests. Three kinds of content are exempt: anything ‘evidently fictional’; parody, satire, commentary or criticism; and content ‘limited to text’, so AI-written ad copy needs no label. That last point sets it apart from Europe, where AI Act transparency rules are pushing [OpenAI to hide a watermark in ChatGPT’s writing](https://madrobot.blog/2026/10/06/openai-chatgpt-text-watermark-eu-ai-act-textgrain/). The law doesn’t define ‘significantly modified’, leaving open whether retouching, an AI-made background or a cleaned-up voiceover counts. A label is needed only when the AI content misleads about the goods or services themselves, and only if the advertiser acted knowingly, recklessly or willfully. Because it is about goods and services, it doesn’t reach political ads. Enforcement is limited too. Only the state attorney general or a district attorney can sue; shoppers can’t. Courts can order an ad stopped and impose a civil fine of no more than $3,000, plus legal fees. The Pennsylvania Chamber of Business and Industry backed the May amendment that set this up, telling senators in a [memo](https://www.pachamber.org/wp-content/uploads/2026/05/MEMO-to-Senate-Comms-Tech-SupportAmend_-SB-806.pdf) that it made the bill ‘significantly more workable for the business community’ by limiting enforcement to prosecutors rather than ‘expansive liability’ under the state’s consumer protection law. The tech industry group NetChoice opposed the bill, arguing in an October 4 letter to lawmakers, [reported by MediaPost](https://www.mediapost.com/publications/article/418601/pennsylvania-passes-bill-requiring-disclosure-of-m.html), that existing laws already ban deceptive ads and that rules ‘this prescriptive risk overwhelming short videos, small display ads, and audio spots’. ## How it compares with other states Most state rules on AI in ads so far are about elections: Public Citizen’s [tracker](https://www.citizen.org/article/tracker-legislation-on-deepfakes-in-elections/) lists more than 30 states with laws on deepfakes in political campaigns. Rules for ordinary commercial ads are rarer. New York’s [‘synthetic performer’ law](https://www.nysenate.gov/legislation/bills/2025/S8420), signed in December 2025 and in force since June, makes advertisers disclose AI-generated people in ads, with fines of $1,000 for a first violation and $5,000 after that. Pennsylvania’s law is broader in one way, covering any AI-made or AI-altered image, video or audio, but bites only when the ad misleads about the product. ## Why it matters AI-generated product photos, voices and video are becoming routine in advertising, and Pennsylvania has now put a price, if a small one, on using them to deceive. With ‘significantly modified’ undefined and fines capped at $3,000, how far the law reaches will depend on how hard the attorney general and district attorneys choose to push it.