ASA/CAP AI advertising claims: what substantiation does an AI ad need?
Last reviewed May 30, 2026
When an AI product ad uses 'AI-powered', 'no judgement', or 'unlimited creative freedom' wording, ASA and CAP apply the same advertising rules that govern any other performance claim: the ad must give viewers an accurate and complete impression of what the product actually does. This page shows what function detail, content restrictions, and audience safeguards an AI ad should include—and what evidence to request when reviewing an AI product through its advertising.
Evidence buyers verify
- A plain-language description of what the AI actually does—input, processing, and user-visible output.
- Visible content restrictions or prohibited-use categories in the same ad context as any capability or freedom claim.
- Audience controls, age targeting, and prohibited-scenario disclosure for any companion, roleplay, or social AI product.
Opens the checker for this claim type. Paste your vendor's exact wording there. Evidence questions only — not a blacklist or fraud detector. Not sure what a result looks like? See a sample receipt.
Sources this guide draws from
- · November 2024
ASA analysis of 16,000 UK ads using 'AI' as a descriptor. Found that many ads do not explain what the AI does, its input, output, or user-visible benefit. Advised advertisers to avoid implying capabilities that do not exist.
- · 2026
ASA upheld complaint against Aurai AI roleplay app ad. Companion-style positioning appeared alongside content implying scenarios without consent. Breached CAP Code rules 1.3 (Social responsibility) and 4.1, 4.4 (Harm and offence).
- · 2024
ASA/CAP guidance clarifying that existing advertising rules apply regardless of whether AI is used, and that 'disclosure alone is very unlikely to mitigate the harm caused by a fundamentally misleading message'.
Public claims with documented evidence gaps
"No judgement, no pressure – just a friend who gets you"
Vague AI-powered- Source and date
- ASA ruling: Haikou Chengfa Technology Co Ltd (Aurai AI) · 2026
- Evidence signal
- Companion-style positioning that implies no content limits, no audience controls, and unrestricted interaction—without visible restriction or safeguard disclosure.
- Evidence gap
- ASA found the 'no judgement' companion positioning appeared without visible content controls, prohibited-scenario disclosure, or audience-age safeguards. The ad impression created did not match the restrictions that should apply to an AI roleplay product.
- Buyer question
- What content restrictions and prohibited scenarios apply to the AI companion, and are those limits visible in the same ad context as the 'no judgement' claim?
Match each claim pattern to the evidence buyers need
| Claim pattern | Evidence needed | Buyer question |
|---|---|---|
| AI-powered, AI-driven, or AI-enabled used without function detail | Specific AI function, input, output, user-visible benefit, human oversight boundary, and a plain-language description of what the product does that a viewer could verify. | If the words 'AI-powered' are removed from the ad, what concrete product capability remains, and where does the ad describe it? |
| AI companion or no-judgement positioning | Audience controls, age targeting, prohibited interaction scenarios, content moderation boundary, complaint handling process, and whether the same restrictions are visible in the ad context. | What content restrictions and prohibited scenarios apply, and are they as visible in the ad as the 'no judgement' or 'always available' wording? |
| Creative freedom or no-limits claims for AI image or video tools | Allowed-use policy, prohibited content categories, consent rules for real-person likenesses, automated blocking scope, and whether the ad creative matches the restrictions users actually face. | What content does the product block, and are those limits visible in the same marketing context as the creative-freedom claim? |
| AI product benefit claim in short-form or in-app ads | Plain-language benefit statement, evidence that AI produces that benefit in the user workflow, product restrictions, material limitations, and whether the ad shows the same output a user would receive. | What evidence shows the advertised benefit comes from the AI in the product workflow, not from edited demos or unrepresentative creative presentation? |
| AI accuracy, performance, or quality claims in ads | Test scope, sample or benchmark used, conditions tested, excluded use cases, and whether the ad's claimed result matches typical user experience rather than a best-case demo. | What test supports the accuracy or quality claim, and do the test conditions match the inputs and tasks a typical user would rely on? |
Evidence to request
- A plain-language description of what the AI actually does—input, processing, and user-visible output.
- Visible content restrictions or prohibited-use categories in the same ad context as any capability or freedom claim.
- Audience controls, age targeting, and prohibited-scenario disclosure for any companion, roleplay, or social AI product.
- Evidence that the ad's claimed benefit comes from the AI in the actual user workflow, not from demo editing or creative presentation.
- ASA/CAP rule applicability: whether the ad is subject to CAP Code (non-broadcast) or BCAP Code (broadcast) and which rules apply to the claim type.
Questions to put in front of the vendor
- What AI function, input, and output does the product actually perform, and where does the ad describe this in plain language?
- What content restrictions or prohibited scenarios apply to the product, and are they visible in the same ad context as the benefit or freedom claim?
- For companion or roleplay AI products, what audience age controls and prohibited-interaction safeguards are in place and publicly documented?
- Does the ad show the output a typical user would receive, or a best-case edited demo?
- If the ad claims AI generates or improves content, what evidence shows the AI produces that result—not post-production editing or cherry-picked examples?
- Is the ad placed in a context where audience-age controls apply, and what evidence shows the placement matches those controls?
Wording boundaries to compare against
- AI [function]: takes [input], produces [output]; content restrictions: [named prohibited categories]; audience: [age gate or targeting controls].
- AI companion that [specific function]; prohibited scenarios: [list]; content moderation: [automated blocking scope + human review boundary].
- AI creative tool: allowed uses include [list]; prohibited content: [list]; consent requirement: [yes/no] for [named content type].
- AI-generated [output type] based on [input]; typical result range: [description]; best-case result shown in ad was produced under [conditions].
Have your vendor's exact claim wording ready?
Check an AI advertising claim How the evidence method works