AI Ad Disclosure: How Should AI-Generated and In-Chat Ads Be Labeled?

AI Ad Disclosure Case 1: Ads Made With AI

When AI helps create an ad, the question is whether the audience would be misled without being told. The IAB's AI Transparency & Disclosure Framework answers with a materiality test: disclose when AI involvement materially affects authenticity, identity, or representation. Our explainer on the framework covers the detail. In practice:

• Disclose: a synthetic spokesperson who looks like a real customer, a cloned or synthetic voice, a digital twin of a real person, or a realistic scene that never happened and could be mistaken for documentary footage.
• Usually no label needed: AI used for routine editing, resizing, background cleanup, or draft copy that a human then reviewed and changed.
• Check the law: some synthetic performer cases now carry legal duties, for example New York's rule for synthetic performers in ads.

AI Ad Disclosure Case 2: Ads Inside AI Conversations

The second case is different. Here the ad is not necessarily made with AI. It is delivered inside a conversation with an AI, right beside the AI's own answer. The risk is that the person cannot tell where the assistant's advice ends and a paid message begins. Industry frameworks do not yet define a dedicated spec for this placement, so good practice matters more than a checklist.

Effective disclosure for in-conversation ads has four properties. It is visible, meaning it is not hidden in small type. It is plain, using the word "Sponsored" or "Ad" and not a coined label. It is adjacent, sitting on or directly next to the ad itself. And it is consistent, shown the same way every time so people learn to recognize it.

Examples of Clear AI Ad Disclosure Wording

One more point applies to both cases: disclosure is a trust decision as much as a compliance one. People are far more forgiving of AI involvement that is stated openly than of AI involvement they discover later. An honest label costs a few words and often goes unnoticed. A missing one, once spotted, can cost a brand credibility with the very audience it was trying to win.

A Simple Decision Process for AI Ad Disclosure

• Is the ad paid? If yes, label it as sponsored wherever it appears.
• Does it include a synthetic person, voice, or realistic scene? If yes, add an AI-content disclosure and check applicable law.
• Could a reasonable person mistake it for the assistant's own advice? If yes, make the separation stronger.
• Does it appear in a sensitive conversation? If yes, don't run it.
• Can you show a regulator or a client exactly what was disclosed? If not, add record-keeping.

Common AI Ad Disclosure Mistakes

• Treating the two cases as one. An in-chat sponsored label does not cover a synthetic spokesperson inside the creative, and the reverse is also true.
• Over-labeling. Tagging every lightly AI-assisted asset trains audiences to ignore the label that matters.
• Hiding the label. Small type, low contrast, or placement behind a click is not clear disclosure.
• Coined terms. Phrases like "powered by partners" do not tell a person they are looking at an ad.
• No records. If a client or regulator asks what was disclosed on a given date, you need to be able to show it.

Who Should Own AI Ad Disclosure Inside a Company

Disclosure often falls between teams. Creative produces the assets, media buys the placements, and legal reviews the risk, so no one owns the whole. A workable pattern is to give one person responsibility for a short written policy that covers three things: which AI uses require a label, the standard wording for each, and where records are kept. Review it each quarter, because guidance from IAB, the EU, and US states is still developing. Agencies should agree on the policy in writing with clients before a campaign starts, not after.

How Thrad Approaches AI Ad Disclosure

At Thrad, every placement is clearly labeled as sponsored, and ads are withheld from sensitive conversations. That reflects the same principle behind the IAB's framework: tell people what they need to know, and don't run ads where they do not belong. The full practice is on the safety page.

Disclosure Tips for Agencies and In-House Teams

• Add a field to the creative brief that asks whether any AI tool produced a person, voice, or realistic scene.
• Keep a one-page reference with approved label wording for each ad format you run.
• Ask platforms to confirm in writing how sponsored content is labeled on their surfaces.
• Screenshot live placements at launch, so you hold proof of what the audience saw.

What Will Change

Expect more specific guidance. IAB is holding a workshop on its framework on October 28, 2026, and regulators in the US and EU are still issuing guidance on the new rules. Until a dedicated standard exists for ads inside AI conversations, disclosing more clearly than the minimum is the safer path. For the full picture of how brands and publishers handle this, start with our pillar guide to conversational advertising.