The Interactive Advertising Bureau has rewritten the rulebook for when an ad has to admit a machine made it. Version 2 of the IAB’s AI Transparency and Disclosure Framework, released August 18, replaces a one-size-fits-all instinct toward labeling with a narrower test: disclose only when AI use could genuinely mislead a consumer about authenticity, identity, or representation. For marketing leaders who spent the past year guessing at what regulators and platforms would eventually require, the guesswork just got a documented answer, even if it is not the strict one privacy advocates wanted.
What changed
The original framework, published in January 2026, set out the industry’s first attempt at AI disclosure norms. Version 2 does not scrap that structure so much as sharpen it in response to what IAB calls regulatory momentum in New York, California, South Korea and the European Union, jurisdictions that have each moved on AI-content rules in the months since. The updated guidance is explicit about where the line sits: realistic AI-generated images and video, synthetic voices, digital twins, and chatbots that could be mistaken for a human all require disclosure. Routine post-production work, copy drafting assisted by AI, and generic synthetic voices do not.
“Trust is everything between a brand and its customers, and being honest about AI is part of earning it,” said Caroline Giegerich, VP of AI at IAB. She was also direct about why the framework stops short of blanket labeling: “Under-disclosure leaves consumers at risk of being misled. Over-disclosure could risk negatively impacting advertisers.” David Cohen, IAB’s CEO, framed the update in broader terms, calling trust “a foundational element which is critical to the growth of AI across the ecosystem.”
The case against labeling everything
IAB’s argument for restraint rests partly on consumer research it commissioned: more than half of consumers say they want disclosure when an ad is fully AI-generated or uses AI imagery or video, and 73% of Gen Z and Millennial respondents said clear disclosure would increase or have no effect on their likelihood to purchase. Read one way, that data supports mandatory labeling. IAB reads it the other way: since most consumers are not hostile to AI-assisted ads once told, the industry does not need to over-disclose routine production work that carries no deception risk.
There is a harder number behind that judgment call. Research from NYU Stern, reported by PPC Land, found that disclosing generative AI involvement in an ad cut its click-through rate by 31.5%. IAB’s framework treats that decline as the real cost of transparency, one worth paying when a consumer could otherwise be deceived, and not worth paying when they could not be.
The framework’s working group included Acxiom, Mondelez International, Tinuiti and The Weather Company, giving it input from both the brand and agency side of the disclosure question, not just platforms with a stake in ad performance.
What it means for the marketing leader
The practical effect is a compliance test that finally has documented edges. A CMO who has been asking legal and creative teams to guess at disclosure now has a materiality standard to point to: does the AI use affect authenticity, identity, or representation in a way a reasonable consumer would want to know about? If yes, label it. If the AI touched color correction, a script draft, or a stock synthetic voice with no attempt to impersonate a real person, it does not need a label under this framework.
That said, IAB’s framework is guidance, not law. New York, California, South Korea and the EU are still writing their own binding rules, and IAB’s risk-based approach could turn out to be more permissive than what regulators ultimately require. Marketing leaders should treat V2 as a floor for internal policy, not a ceiling, and keep it under review as those jurisdictions finalize their own AI-content statutes. The 31.5% click-through cost also means finance and legal will increasingly be in the same room when a campaign decides whether to disclose, since the framework has now put a number on what compliance can cost in performance terms.
For teams building AI use into campaigns at scale, this also changes vendor conversations. Procurement questions about AI provenance and disclosure readiness are no longer hypothetical; they map directly onto IAB’s four disclosure triggers, which is the kind of specificity legal teams can actually build a checklist around.
Channel by channel, the standard bites differently
The four triggers do not land evenly across formats. A synthetic voice reading a radio spot is a clear disclosure case. A retail brand using an AI-generated hero image in a paid social carousel is another. But a conversational shopping assistant answering product questions inside a chat interface sits closer to the “chatbot that could be mistaken for a human” trigger than most marketers have been treating it, which means the fastest-growing part of the ad stack, agentic and conversational commerce, is also the part most exposed to this update. Programmatic buyers running creative through AI-assisted production pipelines will need clearer documentation from their DSPs and creative vendors about which assets cross the materiality line and which do not.
Enforcement risk is uneven too. A brand operating only in US states without AI-content statutes has more room to lean on IAB’s judgment than one running the same creative into New York, California, South Korea or the EU, where a regulator’s own definition of “material” may not match IAB’s. Global campaigns will likely need to disclose to the strictest jurisdiction’s standard rather than run market-specific creative variants, which is its own cost the framework does not price in.
Where this fits in a busier disclosure landscape
V2 arrives alongside a broader tightening across the ad-tech stack. MarTech Edition has covered the parallel rewrite of ad tech’s privacy rulebook, and the industry’s push toward standardized measurement for AI visibility, detailed in our coverage of the shared AI visibility standard, is part of the same pattern: as AI becomes infrastructure rather than novelty, the industry is building the compliance and measurement scaffolding to match. Disclosure, privacy and visibility are converging into a single governance conversation for anyone running ads.
The near-term test for V2 is whether platforms and agencies actually adopt its four-trigger standard consistently, or whether each treats “materiality” as license to interpret the rule in whatever way suits their own AI tooling. IAB has given the industry a shared vocabulary. Whether that vocabulary becomes a shared practice is the part regulators, and marketers watching their own click-through rates, will be judging next.
Source: IAB via PR Newswire
