IAB updates ad disclosure standards for AI
▼ Summary
– IAB released Version 2 of its AI Transparency and Disclosure Framework, updating guidance on when AI use in advertising requires consumer disclosure.
– Disclosure is required when AI materially affects authenticity, identity, or representation, but not for every AI use; routine post-production and internal workflows don’t automatically need labels.
– The update follows new disclosure laws in California, New York, South Korea, and the EU, and the framework offers a common approach across these differing regulations.
– Consumer research found over half wanted disclosure for fully AI-generated ads or AI-generated imagery/video, supporting targeted disclosure over universal labeling.
– The framework recommends a sparkle icon or clear text for U.S. ads, while the EU’s Article 50 requires disclosure but doesn’t mandate a specific icon.
The Interactive Advertising Bureau has rolled out the second edition of its AI Transparency and Disclosure Framework, offering updated direction on when brands, agencies, publishers, platforms, and tech vendors should inform consumers about the use of artificial intelligence in ads and marketing materials.
At the core of the new guidance is a clear distinction: AI that merely assists in content production versus AI that changes what audiences might reasonably perceive as real. The IAB mandates disclosure when AI materially impacts authenticity, identity, or representation, but stops short of requiring labels for every instance of AI involvement.
That distinction now carries heavier legal significance. Since the framework’s initial release in January, disclosure mandates have taken effect across California, New York, South Korea, and the European Union. The IAB positions its updated framework as a unified industry standard for navigating these varied and sometimes conflicting requirements.
What triggers disclosure
Realistic synthetic content that could shape a consumer’s interpretation of what they are viewing generally calls for a label. This includes AI-generated images and videos from text prompts, certain synthetic voices and avatars, digital recreations of deceased individuals, and digital twins of living people placed in fabricated scenarios that go beyond standard celebrity endorsements.
Conversational AI falls under the same rule. Chatbots and virtual assistants must identify themselves as AI when there is a reasonable chance users could confuse them with human agents.
The framework treats behind-the-scenes AI differently. Routine post-production work, internal operational workflows, copywriting, standard audio cleanup, background music, generic synthetic voices, and obviously animated or stylized characters do not automatically require disclosure. Authorized synthetic voices and digital twins used in conventional endorsement deals also sit outside the automatic-label category.
This approach shifts the focus from whether AI was used to how it was used. A marketer relying on generative AI to draft ad copy faces a far different disclosure calculation than one producing a lifelike video of a person engaging in events that never occurred.
Consumer sentiment supports selective labeling
The IAB’s position is informed by consumer research conducted with Sonata Insights during the original framework’s development. Findings revealed divided attitudes toward AI in advertising. Some consumers welcomed its role in creative production, while others viewed it as lacking authenticity.
Over half of respondents wanted brands to disclose when an ad was entirely AI-generated or contained AI-created imagery or video. That result bolsters the case for targeted disclosure around synthetic content, without suggesting widespread consumer appetite for labeling every AI application in marketing.
Navigating a fragmented regulatory landscape
The framework offers practical guidance for marketers running identical campaigns across multiple jurisdictions with differing rules. New York’s synthetic performer legislation took effect in June, while California’s SB 942 and Article 50 of the EU AI Act became enforceable on Aug. 2. South Korea introduced its own AI labeling requirements earlier in the year.
For U. S. advertisers, the IAB recommends two disclosure mechanisms: a standardized sparkle icon or plain-text labels. These are industry recommendations, not legal mandates, and do not supersede applicable regulations.
The EU route differs. Article 50 requires disclosure for covered AI-generated content and deepfakes but prescribes no specific icon. A voluntary Code of Practice offers guidance on disclosure design and placement, while the question of a common EU icon remains unresolved.
The expanding patchwork of rules turns AI disclosure into yet another classification challenge for marketing operations. Knowing that AI touched an asset no longer suffices. Companies must increasingly track what AI actually did, whether it altered the authenticity or identity depicted in the content, where that content will run, and which disclosure requirement applies in each context.
(Source: MarTech)
