IAB Releases AI Disclosure Framework V2: The Marketer’s Action Plan

On August 18, 2026, the Interactive Advertising Bureau (IAB) published Version 2 of its AI Transparency and Disclosure Framework — the advertising industry’s most comprehensive self-regulatory guide to date for how brands, agencies, publishers, and platforms should handle AI in consumer-facing advertising and marketing. The update reshapes the default question from “does AI touch this?” to “does AI change what a consumer reasonably believes is real?” That shift has immediate, practical consequences for every marketing team using generative tools.

This article breaks down exactly what V2 requires, what it exempts, who it applies to, and the five concrete steps you should take before your next AI-assisted campaign goes live.

What is the IAB AI Transparency and Disclosure Framework V2?

The IAB’s framework is a voluntary industry standard — not a regulation — that gives advertisers, agencies, publishers, ad tech platforms, and measurement partners a shared baseline for when and how to disclose the use of AI in advertising. Version 2, released August 18, 2026, builds on the original 2025 framework and introduces a significantly more nuanced approach: a risk-based, materiality-driven model that avoids the blanket “label everything AI” instinct many brands defaulted to after generative tools became mainstream.

The core principle is that not every AI-assisted creative asset needs a consumer-facing label. Requiring disclosure on routine production tools — background color correction, grammar checking, automated image resizing — trains audiences to ignore labels entirely, which ultimately undermines the trust disclosure is supposed to build. V2 reserves the label requirement for AI uses that could genuinely mislead consumers about what is authentic, real, or who is speaking.

What changed from V1 to V2?

Version 1, published in 2025, established the principle that AI involvement in advertising should be transparent. Version 2 operationalizes that principle with clearer decision criteria. The key changes:

• Materiality threshold introduced. Disclosure is now triggered by whether AI use “materially affects authenticity, identity, or representation in ways that could mislead consumers” — not by the mere presence of AI in the production process.
• Exemption list expanded. Routine production tasks, clearly stylized creative (where the AI aesthetic is obviously non-realistic), and background optimization tools are explicitly exempted from consumer-facing disclosure.
• Shared baseline across the supply chain. V2 extends obligations to all parties in the ad supply chain — not only the advertiser. Agencies, publishers, ad platforms, and data vendors each have defined responsibilities depending on where AI is applied.
• Jurisdiction-agnostic framing. Rather than mapping to a single market’s regulations (EU AI Act, FTC guidance, etc.), the framework is designed as a floor that works across markets, giving global brands a single internal policy they can apply without jurisdiction-by-jurisdiction re-engineering.

The materiality test: when does AI require a label?

The materiality test is the practical heart of V2. Before any AI-assisted asset goes to market, the team responsible for it should ask one question: “If a consumer knew AI generated or altered this element, would it change how they perceive the authenticity, identity, or intent of the ad?” If the honest answer is yes, disclosure is required.

AI uses that require consumer-facing disclosure

• AI-generated or AI-altered human faces, voices, or likenesses that represent real or realistic people, including synthetic spokespersons who appear human.
• AI-generated testimonials or endorsements that imply a real person’s firsthand experience, especially in health, finance, or other regulated categories.
• AI-created scenes presented as documentary footage — for example, a generated “before and after” result that looks photographic.
• AI-written personalized messages framed as being authored by a named human (a CEO statement, a personal recommendation from a named agent).
• Deepfake-style voice or video where a real person’s likeness is replicated without explicit disclosure of the synthetic nature.

AI uses that do NOT require consumer-facing disclosure

• Background retouching, color grading, or image resizing using AI tools.
• AI-powered copyediting, grammar correction, or headline A/B optimization.
• Clearly stylized or obviously non-photorealistic AI art where the synthetic origin is visually self-evident.
• Algorithmic audience targeting, bid optimization, or programmatic placement (infrastructure AI, not creative AI).
• AI-generated translations of existing approved copy.
• Dynamic creative optimization that swaps text or product images within a pre-approved template.

The distinction matters. According to the IAB’s own guidance, “labeling everything teaches consumers to ignore labels and could negatively impact advertisers.” Indiscriminate disclosure creates noise that drowns out the signal — the cases where consumers genuinely need to know.

Who the framework applies to — and what each party must do

V2 maps obligations across the ad supply chain. If your organization plays any of these roles, you have a defined responsibility:

Role	Primary obligation under V2
Advertiser / Brand	Set AI use policy; require agency and platform partners to disclose AI applications that meet the materiality threshold; include disclosure requirements in contracts.
Creative Agency	Document which production elements used generative AI; surface material AI use to the brand before delivery; apply required labels or disclosures to applicable assets.
Media Agency / Buying Platform	Flag AI-driven targeting and optimization tools to clients; ensure publisher-side AI tools used in delivery are documented and accessible for audit.
Publisher	Disclose AI use in editorial content that runs alongside advertising; apply labels when AI-generated native ad content could be confused with editorial or user-generated content.
Ad Tech / Data Vendor	Provide clients with documentation of where AI is applied in the data pipeline; enable reporting that lets buyers distinguish AI-driven from non-AI-driven signals.

5 steps to build your AI disclosure policy before your next campaign

The framework is voluntary, but brands that act now build institutional processes before regulators or litigation force a reactive scramble. Here is a practical sequence:

Step 1: Audit your current AI use in marketing

Before you can disclose, you need to know what you are using. Map every tool in your marketing stack that has an AI or generative component — creative production, copy generation, personalization engines, ad optimization, social scheduling, analytics. Note which produce consumer-facing output and which operate in the background.

Step 2: Apply the materiality test to each use case

For each AI touch-point identified above, run the materiality question: would a consumer’s perception of authenticity, identity, or intent change if they knew AI was involved? Tag each use case as Disclose, Exempt, or Review. Borderline cases should go to legal or compliance for a final call.

Step 3: Define your disclosure format and placement

The IAB framework does not mandate a specific label format, but consistency matters for brand credibility and for building consumer literacy. Common approaches include a brief “AI-generated imagery” label in the ad unit, a disclosure in the footer or terms of an email, or a tooltip in interactive formats. Whatever format you choose, it should be visible, plain-language, and placed where consumers can act on it before engaging with the content.

Step 4: Update agency and platform contracts

Require partner agencies and technology vendors to document AI use and to alert you when that use meets the materiality threshold. This is not punitive — it is supply-chain hygiene. Brands that do not ask will not know, and regulatory liability does not stop at the agency door.

Step 5: Train the people who create and approve content

Frameworks fail when they exist only as PDFs. Run a practical session with your creative, social, and paid media teams that walks through real examples — what counts as a synthetic face, what counts as a testimonial, what a compliant label looks like. The materiality test is a judgment call; teams that practice it consistently make better calls faster.

How V2 fits with platform rules and government regulations

The IAB framework is designed to complement — not replace — platform-specific policies and emerging government regulations. A few things to keep in mind:

• Platform rules vary. Meta, Google, and TikTok each have their own AI content labeling requirements for ads. Some are stricter than V2, some narrower. Where platform rules exceed IAB guidance, follow the platform.
• FTC guidance is evolving. The U.S. Federal Trade Commission has signaled interest in AI-generated endorsements and testimonials as a deception risk under existing rules. V2’s endorsement clause aligns with that risk area.
• EU AI Act obligations overlap. For brands active in Europe, the EU AI Act creates legally binding transparency requirements for certain AI systems. V2 is jurisdiction-agnostic, but it covers substantially the same ground for advertising-specific use cases.

The practical implication: treat V2 as your internal policy floor. Overlay platform rules on a platform-by-platform basis. Use legal counsel for jurisdiction-specific obligations where material penalties attach.

Why this matters for your brand’s AI content strategy

The timing of V2 is not accidental. Research consistently shows that consumer trust in AI-generated content is declining even as usage rises. At the same time, E-E-A-T signals — experience, expertise, authoritativeness, and trustworthiness — have become the primary differentiator for brands that want to earn visibility in both traditional and AI-mediated search. Disclosing AI use where it matters, and not disclosing where it does not, is precisely the kind of credibility signal that builds long-term audience trust.

Meanwhile, Google’s scaled content abuse policy specifically targets AI content that appears designed to manipulate rankings rather than serve readers. Brands that embed genuine transparency practices into their AI workflows are better positioned on both fronts: they satisfy the spirit of platform policies, and they build the authentic authority that earns citations in AI search.

The bottom line

The IAB AI Transparency and Disclosure Framework V2, released August 18, 2026, gives the advertising industry a workable, risk-based answer to one of its most contested questions: when does using AI in marketing require telling consumers? The answer is not “always” and it is not “never” — it is “when it would materially change what a consumer believes is real.” That standard is implementable, defensible, and aligned with the direction regulators in the US and EU are already moving. Brands that build V2 into their creative and contracting workflows now will be ahead of the compliance curve, not scrambling to catch up to it.

Frequently Asked Questions

What is the IAB AI Disclosure Framework V2?

Version 2 of the IAB AI Transparency and Disclosure Framework, released August 18, 2026, is a voluntary industry standard that tells advertisers, agencies, publishers, and ad tech platforms when and how to disclose AI involvement in consumer-facing advertising. It introduces a materiality-driven test: disclosure is required only when AI use could mislead consumers about authenticity, identity, or who is speaking.

Does every AI-generated ad need a disclosure label?

No. Version 2 explicitly exempts routine production AI — background retouching, grammar tools, dynamic creative optimization, and audience targeting algorithms — from consumer-facing labels. Disclosure is required for AI uses that affect authenticity or identity, such as synthetic human voices, AI-generated faces presented as real, or AI-written testimonials attributed to real people.

Does AI-written ad copy need an AI disclosure?

Not automatically. If the copy is part of a standard campaign ad and does not misrepresent who authored it or imply a specific human’s firsthand experience, V2 does not require a label. If the copy is framed as a personal endorsement from a named individual who did not write it, that is a different scenario and likely requires disclosure.

Is the IAB framework legally binding?

No, it is a voluntary self-regulatory standard. However, it provides a defensible compliance baseline that aligns with FTC guidance on AI endorsements and overlaps significantly with EU AI Act transparency obligations for advertising use cases. Brands that follow V2 are better positioned if regulators or litigation scrutiny arrives.

How is V2 different from platform AI labeling rules?

Platform rules (Meta, Google, TikTok) are platform-specific, mandatory within each ecosystem, and vary in scope. The IAB framework is cross-platform, voluntary, and covers the full ad supply chain — including agency and vendor relationships that platforms do not see. Where a platform rule is stricter than V2, follow the platform rule.