Aipedia: Marketing and PR agencies

AI ad rules for agencies: what the ASA, CMA and IPA require

Aipedia, by the Apex broking team · Last reviewed 2026-10-27

No UK rule bans or specifically regulates AI in ads; the CAP Code, the DMCC Act and data law apply to AI-made content exactly as they apply to anything else. For agencies, that means the tool is never an excuse when a claim misleads or a review is fake.

Part of: AI in Marketing and PR agencies

In short

There is no AI-specific rule in the CAP or BCAP Codes; the ASA applies its media-neutral rules to ads made with AI and reports rising complaints. The DMCC Act bans misleading actions (s.226) and fake or concealed incentivised reviews, including offering review services (Sch 20 para 13), from 6 April 2025. CMA207 and CMA208 explain the law but do not address AI. The voluntary IPA/ISBA 12 principles expect openness where AI is prominent, no undisclosed deepfakes, human oversight and client agreement. The CIPR added PR guidance in June 2026.

Who sets the rules for AI-made marketing

Marketing and PR agencies have no statutory regulator of their own. Instead, the work they produce is judged by several bodies, each with a different status. The ASA enforces the advertising codes written by CAP (non-broadcast) and BCAP (broadcast). The CMA enforces consumer protection law under Part 4 of the Digital Markets, Competition and Consumers Act 2024 (DMCC Act). The ICO enforces UK GDPR and PECR for data and direct marketing. Alongside these, trade and professional bodies publish voluntary principles: the IPA for agencies, ISBA for advertisers, and the CIPR for PR practitioners.

None of these bodies has created a separate regime for AI. The practical question for an agency is therefore how existing rules apply when a model writes the copy, generates the image or drafts the review.

At a glance

Body and publication	Status	Date	Key point on AI
ASA/CAP, CAP and BCAP Codes	Binding self-regulation	Ongoing	No AI-specific rule; media-neutral rules apply to AI ads
ASA and CAP 2025 Annual Report	Report	2026	Rising AI ad complaints; CAP considering AI capability claims
CAP News on misleading imagery	Guidance	11 September 2025	Altered images must not exaggerate efficacy; does not mention AI
DMCC Act s.226	Law	In force 6 April 2025	Misleading actions, including overall presentation
DMCC Act Sch 20 para 13	Law	In force 6 April 2025	Fake and concealed incentivised reviews banned
CMA207 and CMA208	Statutory guidance	4 April 2025	Explain the rules; neither expressly addresses AI
IPA/ISBA 12 principles	Voluntary	1 November 2023, updated 6 March 2025	Disclosure, IP, oversight and client consent
Online Advertising Taskforce best practice guide 2026	Voluntary	Updated 5 February 2026	Risk-based disclosure and oversight decisions
CIPR, The Responsible Use of AI in PR	Voluntary, members only	June 2026	Translates the Venice Pledge into PR practice

ASA and CAP: same rules, more complaints

There is no AI-specific rule in the CAP Code or the BCAP Code. The ASA and CAP 2025 Annual Report explains that CAP's rules are media-neutral, so they apply the same standards to every ad, including ads made with AI. An AI-generated visual is assessed on whether it misleads, causes harm or serious offence, exactly as a photograph would be. The report notes a growing number of complaints about AI-generated ad content in 2025 and says CAP is considering how consumers interpret claims about AI capabilities. That second point matters for agencies promoting AI-powered products: claims about what a product's AI can do will need evidence like any other performance claim.

The ASA is also using AI itself. Its Active Ad Monitoring system processed nearly 60 million ads in 2025, so non-compliant AI content may be spotted proactively, not only on complaint.

The most relevant guidance on visuals is CAP News, A picture says a thousand words: avoiding misleading imagery in ads (11 September 2025). It says images altered by pre- or post-production techniques, including filters, must not exaggerate what a product can do, and cites the Tanologist and Cinzia Baylis Zullo filter ruling. It does not mention AI, but the principle carries straight across to AI-enhanced or AI-generated product shots. See AI-generated images in ads.

CMA and the DMCC Act

Section 226 of the DMCC Act prohibits misleading actions. It covers false or misleading information and an overall presentation likely to deceive the average consumer, even where every individual statement is true. It also covers failing to follow a code of conduct the trader says it follows, so a client who publicly signs up to responsible AI principles and then ignores them could be exposed.

Schedule 20 paragraph 13 bans submitting, or commissioning someone to submit, fake consumer reviews; concealed incentivised reviews; publishing reviews misleadingly; publishing reviews without reasonable steps to prevent fakes; and, under 13(4), offering services to traders to do any of these. An agency that offers AI review generation is directly in scope. Both provisions came into force on 6 April 2025 under SI 2025/272. The CMA published CMA207 and CMA208 on 4 April 2025; neither expressly addresses AI. See AI-written reviews and the fake review ban.

IPA, ISBA and the Online Advertising Taskforce

The IPA and ISBA 12 principles for the use of generative AI in advertising were published on 1 November 2023 and the page was updated on 6 March 2025. The principles most relevant to day-to-day agency work are:

2: no undisclosed deepfakes or fraudulent ads.
3: be open about AI use where it plays a prominent role and consumers are unlikely to notice.
5: avoid bias.
6: respect personal data rights.
7: consider IP holders.
9: carry out due diligence on tools.
10: keep human oversight, including checking facts and permissions.
11: be transparent with partners, and include no AI-generated content in their materials without agreement.
12: monitor use continually.

Principle 3 is the closest thing to a disclosure rule: it is not a blanket labelling duty but a judgement about prominence and consumer awareness. The Online Advertising Taskforce AI Working Group's best practice guide for 2026, hosted by the IPA and updated on 5 February 2026, supports risk-based decisions on disclosure and oversight. IPA members also have legal resources including an AI checklist, client contract clause guidance, liability and indemnity guidance, and the 2025 IPA/ISBA template framework agreement with AI provisions; see AI clauses in agency client contracts.

CIPR guidance for PR

In June 2026 the CIPR published The Responsible Use of AI in PR, a members-only best practice guide translating the Global Alliance's Venice Pledge, which the CIPR has signed. It also offers an AI and digital ethics playbook. These are professional expectations rather than law, but a client or court assessing whether a PR consultant acted competently may look to them.

What it means in practice and for PI

Because the rules are media-neutral, an agency cannot argue that an AI tool made the misleading claim or the fake review. The client is usually the advertiser in the regulator's eyes, but the client may look to the agency for its losses. No statutory or professional-body PI requirement applies to agencies, and no regulator has said anything about AI and PI. A PI or media liability policy may respond to a negligence, defamation or infringement claim, subject to its terms, but regulatory fines are generally uninsurable; see are AI regulatory fines insurable? and which policy responds. Return to the marketing and PR agencies hub for the full set of guides.

How Apex helps: Apex arranges PI for marketing and PR agencies. A broker can check how your wording treats claims arising from AI-assisted ads, reviews and PR content.

Sources

Frequently asked

Does the CAP Code require AI-generated ads to be labelled?

No CAP or BCAP rule specifically requires AI labelling. An ad must not mislead, so disclosure may be needed where AI use would otherwise give a false impression. IPA/ISBA principle 3 recommends openness where AI plays a prominent role and consumers are unlikely to notice.

Can an agency breach the DMCC fake review ban?

Yes. Schedule 20 paragraph 13(4) bans offering services to traders to submit fake reviews, conceal incentivised reviews or publish reviews misleadingly. An agency offering AI-written reviews or testimonials as a service is directly in scope, not only the client.

Are the IPA/ISBA principles legally binding?

No. They are voluntary industry principles. They still matter because clients may write them into contracts, and they show what a competent agency is expected to do, which can be relevant if a client alleges negligence.

Check your PI covers how you use AI

Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.

Start a PI quote Request a callback

Apex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.