IAB AI Transparency & Disclosure Framework: Complete Guide 2026

Table of Contents

1. What Is the IAB AI Transparency & Disclosure Framework?

The IAB AI Transparency and Disclosure Framework is an industry standard published by the Interactive Advertising Bureau (IAB) in January 2026. It is the advertising industry's most comprehensive attempt to answer one question that every brand, agency, and creative team is now asking: When AI creates or modifies ad content, does the consumer need to know?

The framework does not say "always disclose everything AI." Nor does it say "AI is just another production tool, no disclosure needed." It does something more nuanced — and more useful. It defines a materiality threshold: a test that determines whether AI involvement in content creation is significant enough that a reasonable consumer would care if they knew about it. If the answer is yes, disclosure is required and the framework tells you exactly what to say and where to put it.

The IAB is not a regulator. It cannot fine you. But the framework matters enormously for three reasons:

First, it is rapidly becoming the de facto compliance standard that major platforms, DSPs, and publishers are adopting as their enforcement baseline. When Google, Meta, or The Trade Desk decides how to handle AI disclosure requirements at the platform level, they will build toward this framework.

Second, it maps directly onto the disclosure requirements embedded in binding law — particularly the New York Synthetic Performer Law (effective June 2026), California SB 942 (effective August 2026), and the EU AI Act — giving compliance teams a single framework they can use across jurisdictions.

Third, it defines the C2PA custom assertions that ad tech infrastructure uses to signal AI disclosure compliance in machine-readable metadata. If you are building or buying ad tech in 2026, the IAB framework is the specification you are building to.

2. Why the IAB Created the AI Transparency Framework (January 2026)

The advertising industry's existing disclosure frameworks were not built for this. Prior standards assumed that the humans creating advertising content were humans, using human labor. The question of "did AI make this?" simply was not on the regulatory or ethical checklist.

Three things converged to make the IAB move:

Regulatory Pressure & Real Financial Strings

The legal landscape is no longer hypothetical. Legislators in New York and California have passed laws with sharp teeth and fines upto $5,000.

The New York Synthetic Performer Law and California SB 942 require clear disclosure of AI-generated performers and content. Suddenly, the legal landscape was fragmenting, and the industry needed a unified standard to navigate it without managing 50 different sets of rules.

The IAB's response was the January 2026 framework: the first industry standard that is simultaneously a decision-making guide, a vocabulary of specific label texts, and a technical specification for C2PA metadata.

3. The Three Disclosure Checks

Before requiring a disclosure label, the IAB framework asks three simple yes/no questions. These determine if the AI involvement is significant enough that a consumer needs to know about it. If all three answers are "Yes", a label is required.

The materiality test is intentionally practical — it is designed to let routine AI-assisted production continue without disclosure burdens, while drawing a clear line at the point where AI involvement could actually mislead a consumer about what they are buying or who is endorsing it. Think of the three checks as a filter: most AI post-production work fails Check 1 and never needs a label. Only the cases that clear all three reach the point of required disclosure.

It is worth noting what the test is not asking. It does not ask whether AI was involved at all — the framework explicitly accepts that AI tools are now part of the standard creative production stack. It does not ask whether the brand intended to deceive. It asks only whether the output, as a consumer would perceive it, clears these three thresholds. An AI tool used in good faith with no deceptive intent can still produce content that requires a label if the output meets the materiality criteria.

4. Content-Type Rules: Images, Video, Audio, Synthetic Influencers, Text

The IAB framework applies the materiality test differently across five content types. Each has specific decision branches, specific exemptions, and specific label texts. Understanding which category your creative falls into is the most important step — the downstream disclosure requirements and legal exposure vary significantly across them.

4.1 Images

The most common application of the framework, because AI image generation is now the default production tool for many creative teams. The key variable is photorealism — not the tool used, not whether the image started from a real photograph.

A photorealistic AI render of a coffee cup on a kitchen counter requires disclosure. A clearly illustrated or 3D-animated version of the same coffee cup does not. The distinction is not aesthetic — it is whether the image could pass for a real photograph in a casual viewing context.

The trickiest edge case is substantive editing of real photography. The framework draws a precise line here: AI changes that alter the appearance of the product itself or of a real person shown in the ad require disclosure. Background-only replacements are generally exempt, provided the new background does not misrepresent where or how the product performs. Removing a cluttered shelf and replacing it with a clean studio backdrop is fine. Adding a beach backdrop to a moisturiser to imply it's water-resistant when it isn't is not.

4.2 Video

Video rules follow image logic but add a timing requirement: when a label is required, it must remain visible for a minimum duration so that consumers actually register it.

If disclosure is required, the label must appear for at least 2 seconds at the first appearance of the AI-generated or AI-altered elements — not buried at the end of the spot. A label that flashes for a half-second in a corner position does not satisfy this requirement.

AI upscaling, frame interpolation, and noise reduction are all exempt. These are established post-production techniques that consumers neither expect to see disclosed nor would find material to their understanding of the product. The disclosure obligation activates when AI is generating or altering something a consumer would consider real.

4.3 Audio

Audio rules are built around identity and authenticity, and they carry the sharpest legal exposure of any content type for U.S. advertisers. The framework's audio provisions map directly onto the New York Synthetic Performer Law — the binding legislation with the hardest deadline and the clearest penalty structure.

The deceased performer rule is a hard line with no exceptions. It does not matter whether the estate has consented, whether the estate was paid, or whether the use is respectful of the performer's legacy. Any AI-synthesized voice of a deceased person in a commercial context requires disclosure. This is one of the clearest and most consistently enforced provisions across all jurisdictions — IAB, New York, California, and the EU are aligned on this.

For living persons, the framework introduces a context dependency: authorized voice clones used for scripted commercial copy are generally exempt. The line is crossed when the voice clone is used to generate statements about real-world facts, fabricate endorsements, or create impressions that the person holds views they have not expressed.

Since audio ads often have no visual component, the framework specifies that disclosure must appear in the ad metadata and in any visible companion banner or player interface associated with the ad. For purely audio-only contexts (smart speakers, radio), disclosure in the spoken content or a clear spoken cue is required.

4.4 Synthetic Influencers

This is the highest-risk category for regulatory fines across both NY and California, and the one where brands and agencies are most likely to have compliance gaps they are not aware of. The category is broad — it covers digital twins of real people, fully synthetic AI personas, and interactive AI chatbots appearing in ads.

The digital twin rules for living persons introduce an authorization question that the other content types do not. Using an AI-generated likeness of a real, living person without their explicit authorization is not just a disclosure failure — it is likely a violation of that person's personality rights and potentially a direct breach of the New York Synthetic Performer Law. Authorization is necessary but not sufficient: even with authorization, placing a digital twin of a living person in fabricated events or locations they never visited requires a disclosure label.

Fully synthetic personas — AI-generated people who have never existed — require disclosure if they are realistic in appearance and appear in a primary endorsement role. An AI-generated person in the background of a lifestyle shot is incidental and does not require disclosure. That same AI-generated person speaking directly to camera about a product does. The framework's "primary vs. incidental" distinction is the key variable here.

4.5 Text and Copy

The most lenient category in the current framework, and the one most likely to evolve as AI-generated content becomes more sophisticated.

The framework's reasoning for exempting AI-written copy is grounded in consumer expectations: advertising copy has always been written by professionals using whatever tools were available, and the use of AI in that process does not change what the consumer learns about the product. A headline is a headline — the production method does not affect the consumer's understanding of the claim.

The testimonial exception is important and sits slightly outside the IAB framework proper. Fabricated AI-generated testimonials — fake reviews, manufactured customer quotes — fall under the FTC's general deception standards regardless of whether the IAB framework would otherwise require a label. The IAB framework does not need to address this directly because broader consumer protection law already does.

The framework notes that text disclosure requirements may be revisited as AI-generated testimonials and synthetic social proof become more prevalent. Treat the current text exemption as stable for 2026 but watch for updates.

5. What Are the Official IAB AI Disclosure Labels for 2026?

The framework is prescriptive about both what the label must say and where and for how long it must appear. Standardised label text is not stylistic preference — it is what allows consumers to recognise the labels across different ad contexts, and what allows platforms to detect and enforce compliance programmatically.

The Complete IAB Label Reference Table

The table below maps every content type that requires disclosure to its IAB-specified label text, required placement method, and duration. Click any label text to copy it exactly.

Source: IAB AI Transparency and Disclosure Framework, January 2026, pp. 18–19. Table recreated with additional practical notes.

Three Placement Rules That Most Teams Get Wrong

Label Format Alternatives

The framework accepts three disclosure formats for visual content types. Text labels are the primary method and the recommended default. Two alternatives are permitted when text labels would materially compromise advertising effectiveness.

Platform-applied indicators — such as Meta's "AI info" label — satisfy the IAB requirement when applied automatically based on C2PA metadata, provided they meet IAB conspicuousness standards. The advertiser remains accountable for verifying the platform has actually applied the label. Relying on platform enforcement without that verification is not a compliance strategy.

6. How Does C2PA Automate IAB AI Disclosure Compliance?

One of the most technically significant aspects of the IAB framework is its integration with the Coalition for Content Provenance and Authenticity (C2PA) standard. C2PA is a technical specification for cryptographically signing digital content to establish its provenance — where it came from, what tools were used to create it, and whether it has been altered since publication.

The Custom IAB Assertions

The IAB framework defines two custom C2PA assertions specific to advertising:

• com.iab.threshold — a machine-readable signal indicating whether the AI involvement in this creative has met the IAB materiality disclosure threshold (true/false)
• com.iab.disclosure — a machine-readable signal indicating whether consumer-facing disclosure has been applied (true/false)

These assertions, embedded in a C2PA Content Credentials manifest attached to the ad creative file, allow DSPs, SSPs, and publishers to programmatically verify compliance at the point of impression.

Which Tools Already Write C2PA?

This is a practical question with an uneven answer. As of early 2026:

Tools that write C2PA metadata natively: Adobe Firefly (and the broader Adobe ecosystem via Content Credentials), Leica cameras, certain Microsoft Designer outputs, and a growing number of stock photo platforms.

Tools that do not write C2PA natively (yet): Midjourney, Runway, Sora, ElevenLabs, and most other independent AI creative tools. Content created with these tools requires a separate C2PA signing step to achieve machine-readable compliance.

The practical implication: if your creative workflow relies on tools that do not natively write C2PA, you need a process to add C2PA metadata before the asset enters your ad serving workflow. The IAB framework's adoption of C2PA creates a market signal that will accelerate tool adoption — but in 2026, workflow gaps are common.

7. Which Laws Mandate AI Disclosure in Advertising? (2026 Guide)

Global Comparison Table: AI Disclosure Requirements

New York: Synthetic Performer Law (Effective June 2026)

New York's law is the most specific and most immediately consequential for U.S. advertisers. It requires disclosure when an advertiser uses a "synthetic performer" — an AI replica of a real person's voice, likeness, or performance — in a commercial context. The definition of "synthetic performer" maps closely to the IAB framework's synthetic influencer and audio categories.

Violations carry civil penalties of $1,000 to $5,000 per occurrence. "Per occurrence" is still being interpreted, but the legal consensus suggests it means per ad unit, per flight — making non-compliance at scale a significant financial risk.

The IAB framework's categories and label texts are designed to be compliant with NY requirements. An advertiser following the IAB framework's synthetic influencer and audio rules should be meeting New York's requirements.

California: SB 942 (Effective August 2026)

California's SB 942 requires disclosure of AI-generated content in certain commercial contexts. It is broader in scope than New York's law — applying to a wider range of AI-generated content, not only AI personas — but with somewhat less specificity about what disclosure must look like. Following the IAB framework's label texts satisfies SB 942's requirements for content that falls within both frameworks' scopes.

South Korea: Telecommunications Business Act Amendment

South Korea moved fastest of all the major markets. The amended Telecommunications Business Act, which came into force in early 2026, requires disclosure of AI-generated content in digital advertising. For advertisers running campaigns in South Korea, this is already a live compliance obligation.

EU AI Act

The EU AI Act's provisions are being phased in, with the most relevant commercial obligations coming into full force in 2027. The Act takes a risk-tiered approach and requires disclosure of AI-generated or manipulated media in contexts where it could mislead. The IAB framework is largely compatible with EU requirements, though the EU's obligations also extend into algorithmic transparency in ways the IAB framework does not directly address.

8. The IAB Materiality Checker — Try It Free

Understanding the framework is one thing. Applying it to a specific piece of creative — with its specific content type, specific method of creation, and specific deployment context — requires working through the decision logic.

We built the IAB Materiality Checker to do that in under 90 seconds.

Try the IAB Materiality Checker

The tool does the following:

1. Asks you to select your content type (image, video, audio, synthetic influencer, or text)
2. Asks the relevant follow-up questions for that content type (creation method, editing scope, voice type, etc.)
3. Applies the IAB decision logic to your specific answers
4. Returns a verdict: Disclosure Required, Not Required, or Recommended (context-dependent)
5. If disclosure is required, gives you the exact IAB label text to use and where to place it
6. Shows a jurisdiction breakdown covering IAB, EU AI Act, NY Synthetic Performer Law, California SB 942, and South Korea

9. Who is Responsible for AI Disclosure Compliance?

Advertisers and Brand Teams

You are ultimately responsible for the content you run. If your creative uses AI-generated imagery, AI-synthesized voices, or synthetic personas, you need to know whether disclosure is required. The practical implication: add an AI disclosure check to your pre-flight creative review process. The Materiality Checker takes under 90 seconds per creative asset.

The larger strategic question is governance: who in your organization knows what AI tools are being used in your creative workflow? Many brands are discovering that their agencies or production vendors have integrated AI tools without formal disclosure, leaving the brand holding liability they didn't know they had.

Agencies and Creative Teams

You are producing AI-assisted content at scale, often faster than your compliance review process can keep up. The framework creates a practical need for AI disclosure to become part of the creative production checklist — not a legal afterthought. Teams using Firefly benefit from C2PA being written natively. Teams using Midjourney or Runway need an additional signing step.

The June 2026 New York deadline makes this urgent for any agency with U.S.-based clients running campaigns in New York. "My client didn't ask me to check" is not a defense.

Ad Tech Platforms — DSPs, SSPs, Ad Servers

The IAB framework is a technical specification, not just a policy guide. DSPs and ad servers that accept AI-generated creative need to build toward C2PA manifest verification and should be building toward automated disclosure label enforcement. The com.iab.threshold and com.iab.disclosure assertions are the signals to read.

Publishers that accept third-party ad creative have a platform-safety interest in ensuring that AI-generated content running on their pages is properly labeled. Advertiser non-compliance creates reputational risk for publishers.

Legal and Compliance Teams

The framework's alignment with binding law makes it the starting point for multi-jurisdiction AI compliance. Using the IAB framework as your compliance baseline reduces the risk of fragmented, jurisdiction-by-jurisdiction interpretations that are more likely to create gaps.

10. How to Implement the IAB AI Disclosure Framework: A 6-Step Guide

A practical six-step checklist for getting your organization IAB-compliant.

11. Frequently Asked Questions

12. Related Reading

If this guide raised questions you want to go deeper on, here are the articles in this series that cover each topic in detail.