What Are the ASA & CAP Rules for AI Ads?

The UK Advertising Standards Authority (ASA) and the CAP and BCAP Codes already regulate AI-generated advertising, even though there are currently no AI-specific rules.^[1] Ads that use generative models, deepfakes or other synthetic media are assessed under the existing principles on misleadingness, identification, substantiation and social responsibility, regardless of how content is produced.^[1] Marketers must also consider when disclosure of AI use is needed to avoid misleading consumers, and how synthetic content is likely to be interpreted in real-world contexts.^[2] For a deeper look at disclosure duties, see AI disclosure in advertising.

How existing ASA, CAP and BCAP rules apply to AI-generated ads

The ASA has confirmed that its rules apply “regardless of how content is generated, edited, or targeted”, which explicitly includes AI-generated advertising assets.^[1] The CAP and BCAP Codes require that ads must not materially mislead consumers by action, omission, ambiguity or exaggeration, and these standards apply equally whether content is created by humans, traditional tools or generative AI.^[1] In practice, that means claims about AI products, or visual and audio outcomes depicted using synthetic media, must reflect what typical users can achieve in real life and be supported by evidence.

CAP and BCAP also require that marketing communications are obviously identifiable as ads, for instance through clear labelling and presentation in social or influencer formats.^[3] Where AI tools are used to generate influencer content, personalise creative or simulate user reviews, the obligation to make the commercial nature of the communication clear still rests with the advertiser and any affiliate publisher.^[3] Using AI does not shift responsibility to the technology provider. Marketers remain accountable for ensuring that automated optimisation or synthetic creative does not conceal the promotional intent of the communication.

Misleadingness, substantiation and AI-specific guidance

In guidance on “AI as a marketing term”, CAP emphasises that marketers must not exaggerate what AI products can do and must hold documentary evidence for any performance or capability claims.^[4] The ASA has upheld complaints where advertisers failed to substantiate AI-related claims, for example when an app was presented as delivering outcomes that were not supported by robust data or trials.^[4] Claims that only apply in limited scenarios or to particular users must be clearly qualified, otherwise the ad is likely to be considered misleading under CAP Code rules on qualifications and material information.^[4]

On disclosure of AI involvement, CAP’s position is that there is no blanket legal requirement in the UK to label ads as AI-generated, but existing rules can make disclosure necessary where omission would mislead.^[2] CAP recommends asking two questions: whether the audience is likely to be misled if AI use is not disclosed, and whether disclosure clarifies or contradicts the ad’s message.^[2] The ASA has stressed that simply labelling content as AI-generated will not cure a fundamentally misleading impression, such as an AI-created image showing cosmetic results that are unattainable in practice.^[2] For broader context, see synthetic media in advertising.

Social responsibility, synthetic content and consumer protection

The CAP and BCAP Codes include general social responsibility rules that prevent advertising causing serious or widespread offence, condoning harmful behaviour or exploiting vulnerable groups.^[3] The ASA has indicated that these principles apply directly to AI-generated content, including deepfakes and synthetic voices that may manipulate or distress audiences, particularly children or other protected groups.^[1] Ads using AI to simulate real people, including public figures, must avoid misleading impersonation and respect rules on offence, privacy and consent, as well as sector-specific restrictions such as gambling, alcohol and financial promotions.

Alongside policing industry, the ASA uses AI in its own regulatory operations, including “AI-based filtering” to identify likely breaches at scale, which supports more proactive enforcement.^[5] Its published research agenda signals a focus on high-risk areas such as deepfakes, hyper-personalisation and AI-assisted targeting, especially where vulnerable consumers could be exposed to inappropriate or high-pressure messages.^[5] For broadcast executions, advertisers should coordinate with pre-clearance bodies such as Clearcast, as discussed in pre‑flight Clearcast approvals, to confirm that synthetic audio-visual elements meet BCAP requirements.

Recent ASA practice and what advertisers should do now

ASA casework from 2023 onwards shows growing scrutiny of AI-related claims and imagery, even in the absence of AI-specific rules.^[4] Rulings have focused on unsubstantiated performance promises for AI tools, unclear qualifications and depictions of outcomes that standard users are unlikely to achieve.^[4] The regulator’s own guidance on disclosure of AI in advertising and on AI in regulation signals that this scrutiny will continue as use of synthetic content becomes more routine.^[1]

Advertisers should map AI use across the creative and media workflow, then apply CAP and BCAP rules to each use-case. This includes evidencing AI product claims, stress-testing synthetic visuals and voiceovers for potential to mislead, reviewing targeting strategies for social-responsibility risks, and documenting decisions about whether AI disclosure is needed.^[1] Internal sign-off processes should explicitly include AI checks, so teams can demonstrate that broadcast and online assets comply with current ASA and CAP expectations, even as detailed guidance continues to evolve.^[2]