Is your marketing team using generative AI content in your advertising? On the heels of New York’s and California’s new laws about AI in advertising, Pennsylvania just gave you another reason to pause before you publish.
Gov. Josh Shapiro recently signed into law the Pennsylvania AI Deceptive Advertising Act, adding to the growing state-level focus on AI-generated commercial content. The law takes effect Dec. 21 and could impact your disclosure approach.
The AI Ad Law Landscape
As we’ve shared here, here and here (we’re prolific), Pennsylvania’s law is not the first. Two other states – New York and California – have laid the groundwork when it comes to requiring disclosure of generative AI content in advertising. These laws, supported by the Screen Actors Guild, focus on the use of “synthetic performers” instead of traditional human talent. When an ad includes a realistic human-like figure generated via AI or other technology, disclosures are required in New York and California.
Pennsylvania’s AI Deceptive Advertising Act takes broader aim at the use of any generative AI imagery in ads, if that imagery could deceive consumers. It states: “An advertiser may not knowingly, recklessly or willfully create and disseminate or cause to be created and disseminated an advertisement that uses synthetic content in a manner that creates a false consumer perception about the nature, origin or authenticity of the goods or services being advertised, without clear and conspicuous disclosure.”
“Synthetic content” is defined broadly as “information, including an image, video, audio clip or text, that has been significantly modified or generated by artificial intelligence.”
The Gray Area: When To Disclose
So does this mean that any ad with AI imagery, aka synthetic content, must be labeled as such in Pennsylvania? We’d argue it does not. Instead, a reasonable reading of Pennsylvania’s law indicates that disclosure is required only when the ad’s synthetic content could mislead consumers about the actual products, goods or services offered. Is that footage of a sports car winding its way swiftly down a mountain road AI-generated? If the product being featured for sale is the sports car, then disclose. Is the video a real and accurate portrayal of the sports car’s capabilities, but the sunset behind it is too good to be true and AI-generated? Perhaps no disclosure is required because an AI assist on the sunset will not create a false consumer perception about the nature, origin or authenticity of the sports car, which is the product being sold in the ad.
As noted above, New York and California require disclosure when a synthetic performer appears in an ad. There’s been debate about whether these laws are triggered if the ad only contains a shot of a synthetically generated body part (e.g., a hand washing dishes). This analysis may shift based on Pennsylvania’s law, as a generated body part may not constitute a synthetic performer but will arguably constitute synthetic content. When in doubt, disclose.
How To Disclose
Pennsylvania’s disclosure guidance is fairly prescriptive, requiring disclosures to convey that the ad “contains content that was modified [or generated] using artificial intelligence” and to appear (i) in the first instance when the synthetic content is presented to the consumer; (ii) in the same medium as the synthetic content; and (iii) to the extent technically feasible, in a way that is permanent or extraordinarily difficult to remove from the synthetic content.
Further, the statute defines “clear and conspicuous” disclosure as one that (i) appears in a size, color, contrast, location, duration and audibility that make it readily noticeable, readable and understandable, presented in a language used by a consumer to comprehend synthetic content;(ii) does not contradict and is not inconsistent with other information presented; (iii) appears in close proximity to the claim and unobscured; (iv) if audio, is delivered in a volume/cadence sufficient for a consumer to hear and understand; (v) if visual, is of a size and shade and appears on-screen for a duration sufficient for a consumer to read and comprehend; and (vi) if in print, is in a size, location and color/contrast that make it sufficiently noticeable for a consumer to read and comprehend.
Exceptions
As with New York and California, Pennsylvania includes exemptions, such as for content that is “evidently fictional based on context,” parody, satire, criticism or solely “limited to text.”
Enforcement Risk
Similar to New York’s law, Pennsylvania’s law contains no private right of action; the Pennsylvania state attorney general may seek an injunction or impose civil fines of up to $3,000/violation plus attorneys’ fees. California’s AI law presents greater risk of private enforcement because it deems a lack of disclosure to be an unfair or deceptive act, which might expose advertisers to class actions under the state’s consumer protection laws.
That said, Pennsylvania’s approach to this legislation raises a nuanced enforcement issue: If an ad’s use of an AI-generated image creates a false impression about the products or services advertised, would an AI disclosure cure that? As with most documents written by committee, the language is not crystal clear, so we should consider intent.
We expect Pennsylvania is aiming at basic transparency around material elements and wants consumers to know when products and services featured in ads were created with generative AI. Democratic Sen. Nick Pisciottano, who sponsored Pennsylvania’s AI Deceptive Advertising Act, took to Instagram on Oct. 7, promising this new law would “help regulate AI, and make sure if you see something on the Internet, you know if it’s real or if it was generated with AI. So, we’re leading here in Pennsylvania to make sure there are protections against AI.” Disclosing realistic AI-generated content that materially impacts an ad would seem to meet this goal.
Before You Hit ‘Publish’: An Advertiser Checklist
So what’s an advertiser to do? As generative AI becomes embedded in the creative process, advertisers should consider building AI-specific guardrails and processes into campaign review.
- Know where generative AI is being used. Identify and clearly label for internal purposes any AI-generated or AI-modified elements in copy, imagery, audio, video and other consumer-facing creative. Keep records.
- Ask agencies and vendors. Create contractual obligations and standard operating procedures for outside agencies, production partners and creators to (1) follow your generative AI policy, (2) flag when generative AI is used in campaign assets and (3) include appropriate disclosures.
- Consider the ad holistically. Pay special attention to synthetic performers as well as AI-generated content related to products and services. Consider whether that content may give consumers a misleading impression overall about what they are seeing or hearing.
- Build AI into creative approvals. Add AI use and applicable disclosure requirements to existing advertising review workflows. Include a human review layer prior to publication.
- Think state by state. A national campaign can encounter different AI-advertising disclosure requirements. For national ads, work to create disclosures that meet all relevant state requirements.
- Mind the medium. Complying with state laws may not be enough. Social media platforms have strict guidelines regarding AI content disclosures that go further than these state laws. Broadcast and retail media networks may have their own guidelines related to disclosure of AI content as well.
