California and the European Union will introduce major AI transparency requirements on August 2, while New York has already begun regulating advertisements featuring synthetic performers. Together, the laws create a growing compliance puzzle without establishing a universal standard, leaving many brands to decide for themselves whether disclosing AI use is a legal obligation, a platform requirement or simply the responsible thing to do.
The age of voluntary AI transparency is beginning to close, although not nearly as quickly or cleanly as consumers might expect. On August 2, 2026, two significant sets of AI disclosure rules will take effect, with California regulating large generative AI providers and the European Union imposing transparency obligations on both AI companies and the organizations deploying their technology.
New York has already added another layer to the emerging regulatory patchwork through a law governing synthetic performers in advertising. Yet despite the arrival of all three frameworks, most advertisements created or assisted by AI in the United States will not automatically require a visible disclosure.
That distinction matters because the public conversation has generally treated AI labeling as a single, straightforward issue. The legal reality is considerably messier, with different jurisdictions regulating different technologies, different participants and different forms of synthetic content.
The public conversation has generally treated AI labeling as a single, straightforward issue. The legal reality is considerably messier.
California is primarily concerned with the infrastructure behind generative AI, while the European Union is focused on content that could plausibly be mistaken for something authentic. New York, meanwhile, is targeting an almost opposite scenario by requiring disclosures when advertisements feature entirely invented performers who do not resemble any identifiable real person.
The result is not a unified transparency standard, but three different interpretations of what consumers need to know. For brands and agencies, that means the question is no longer simply whether AI was used, but what kind of AI was involved, what the finished content depicts and where the campaign will be seen.
Three Laws, Three Different Ideas of Transparency
California’s AI Transparency Act, as amended by AB 853, applies directly to providers of publicly accessible generative AI systems with more than 1 million monthly visitors or users. It does not establish a broad obligation requiring every brand or agency using those systems to place an AI label on its advertising.
Covered providers must offer a free public tool capable of helping users determine whether content was generated or modified by their systems. They must also support optional visible disclosures and include mandatory hidden provenance information in AI-generated images, video and audio.
Violations can carry civil penalties of $5,000, with each day of continued noncompliance potentially treated as an additional violation. The law is therefore significant for companies building generative technology, but its direct effect on the average American advertising campaign is much narrower than the headlines might suggest.
The European Union’s Article 50 takes a broader approach by extending transparency responsibilities beyond technology providers. Providers must ensure certain AI-generated or manipulated content can be detected electronically, while professional deployers such as brands and agencies must visibly disclose deepfakes that could be mistaken for authentic material.
The EU definition is not limited to people, celebrities or performers. It can include realistically generated or manipulated representations of people, objects, places, entities and events when viewers might reasonably believe the material is genuine.
Those requirements can apply to advertisers outside Europe when their campaigns reach consumers within the EU. Potential penalties can climb as high as €15 million or 3% of worldwide annual turnover, making geography far less protective than some American companies may assume.
New York’s law approaches the same problem from another direction by regulating advertisements featuring synthetic performers created entirely through artificial intelligence. Rather than focusing on a manipulated representation of someone real, the law covers invented performers who do not resemble an identifiable person.
The requirement has applied since June 9 and follows the location of the audience rather than the advertiser’s headquarters. An advertisement produced and purchased elsewhere can still fall within the law when it reaches consumers in New York, provided the advertiser had actual knowledge that the performer was synthetic.
The first violation can result in a $1,000 penalty, with subsequent violations carrying penalties of $5,000 each. Although those amounts are relatively modest compared with potential EU sanctions, the law introduces another category that brands must identify before launching a campaign.
Most AI-Assisted Advertising Still Falls Outside the Rules
Taken together, the three laws create meaningful obligations without creating a general rule requiring American advertisers to disclose every use of generative AI. An advertisement could use AI for backgrounds, editing, copy development, voice cleanup, visual effects or production assistance and still fall outside all three frameworks.
That gap is likely to become one of the defining tensions in AI advertising. Consumers increasingly want to know when artificial intelligence has shaped what they are seeing, while regulators remain focused on narrower categories involving provenance, deception, synthetic identity or realistic manipulation.
Donatas Smailys, co-founder and CEO of creator video platform Billo, believes the fragmented system will eventually push many brands toward broader disclosure policies of their own. Maintaining separate decision trees for California, New York and the European Union may prove more expensive and operationally risky than applying a consistent standard across campaigns.
“Nobody is going to wait for three different laws to tell them what to do,” Smailys said, arguing that brands are already struggling to determine when a label is legally required. He expects many organizations to disclose by default, particularly as platforms begin identifying and labeling synthetic content before advertisers do it themselves.
That may be the most important practical consequence of the emerging rules. The law is not yet establishing one universal definition of responsible AI use, but the combination of regulation, consumer expectations and platform enforcement is gradually creating one anyway.
Consumers Want More Disclosure Than the Law Requires
Public expectations are already moving well beyond the minimum requirements established by regulators. A second-quarter 2026 survey from Fractl and Search Engine Land, covering 1,008 American consumers and 150 marketers, found overwhelming support for labeling AI-generated media.
According to the research, 91% of respondents wanted AI-generated video labeled, while 90% supported labels for images. Support remained high for audio and written material, at 87% and 84% respectively.
Brand behavior is not keeping pace with those expectations. Only 20% of organizations said they always disclose AI use, while approximately one-third said they never do.
That disconnect creates a reputational risk that cannot be resolved through legal compliance alone. A campaign may be technically permissible without a label while still leaving consumers feeling that something important was concealed.
Smailys said the subject has rapidly become part of the production conversation, even when no law forces brands to ask about it. Six months ago, few advertisers were questioning whether creators had used AI in their videos, but those questions now appear in nearly every brief.
Brands increasingly want to know which elements were captured on camera, which were generated and which were altered after production. That scrutiny reflects a broader realization that AI transparency is becoming part of brand safety, creative governance and consumer trust rather than remaining a specialist legal concern.
Platforms Are Building Their Own Disclosure Systems
Technology platforms are not waiting for lawmakers to agree on a single standard. Google and TikTok have already begun introducing their own systems for identifying, labeling and explaining AI-generated advertising and content.
Google’s My Ad Center includes a “How this ad was made” panel across Search, YouTube and Discover. Advertisements created using Google’s own generative AI tools can receive automatic disclosures, meaning the platform may reveal AI involvement even when the advertiser has not chosen to make it central to the campaign.
TikTok has taken a similarly aggressive approach through its use of C2PA, the industry standard designed to preserve information about the origin and modification history of digital content. The company became the first major video platform to adopt C2PA in May 2024 and now holds a seat on the organization’s Steering Committee.
Through a combination of content credentials, creator-applied labels and invisible watermarking, TikTok says it has labeled more than 3 billion videos as AI-generated. It also removed more than 86 million fake accounts during the first quarter of 2026, underscoring how closely AI transparency is becoming connected to broader questions of authenticity and platform integrity.
These systems remain imperfect, and brands should not assume that every AI-assisted advertisement will be identified automatically. However, they do create a new strategic reality in which the advertiser may no longer control when, where or how the use of AI becomes visible.
The Real Question Is Who Gets to Make the Disclosure First
For American brands, evaluating a campaign now requires more than determining whether someone typed a prompt into a generative AI platform. Teams need to understand which tools were used, what those tools produced, whether the content depicts something plausibly real and where the finished campaign will be distributed.
New York campaigns require particular attention to synthetic people, because the disclosure obligation follows the audience rather than the advertiser’s location. Brands should therefore ask agencies and production partners directly whether anyone appearing on screen was created through AI and document the answer before the campaign launches.
Campaigns reaching European consumers require a different analysis centered on realistic representation. When generated or manipulated material depicts an existing or plausibly existing person, object, place, organization or event in a way that could appear authentic, Article 50 may require a visible disclosure.
California introduces another consideration at the technology level, particularly for companies building or distributing large generative AI systems. Brands using those systems may not be directly regulated by the state’s initial requirements, but they will increasingly encounter embedded watermarks, detection tools and provenance features within the creative supply chain.
The safest operating principle may therefore be broader than any individual law. Brands should develop a consistent internal policy explaining when AI use will be disclosed, how production partners must report it and who is responsible for checking jurisdictional requirements before media goes live.
That does not necessarily mean placing a warning label on every campaign that used an AI-assisted editing tool. It does mean treating transparency as a deliberate brand decision rather than waiting for a regulator, platform or audience to expose the process later.
“Brands that check where they stand under all three laws now will know exactly what to do,” Smailys said, warning that companies avoiding the issue are merely transferring the decision to someone else. A regulator, customer or platform may ultimately answer the disclosure question for them, and those parties may not frame the explanation as generously as the brand would have.
The first generation of AI regulation is not creating a complete transparency system, but it is eliminating the excuse that nobody knows where to begin. Brands can continue treating disclosure as a narrow compliance exercise, or they can recognize that trust is becoming the standard the law has not yet managed to write.