> ## Content Index
> Fetch the complete content index at: https://www.intercepta.ai/blog/llms.txt
> Use this file to discover other available public pages before exploring further.

# AI-generated advertising: who is responsible for each variation
- URL: https://www.intercepta.ai/blog/ai-generated-advertising-asa-ftc/
- Published: 2026-10-29T10:49:19.000Z
- Updated: 2026-10-29T10:49:18.000Z
- Description: An AI tool can return 40 variations from one brief, and a few may alter an approved claim, for example by dropping the conditions of an offer or adding a rating with no evidence supporting it.
- Author: Edward Sweigart
- Tags: Marketing Compliance, AI Compliance, advertising compliance, #rail

*An AI tool can return 40 variations from one brief, and a few may alter an approved claim, for example by dropping the conditions of an offer or adding a rating with no evidence supporting it. In April 2026, the UK’s Advertising Standards Authority (ASA) ruled against two paid search advertisements and their landing page, after identifying the advertisements using intelligence from its own AI monitoring system. The advertiser said one claim had been generated by AI, yet the ASA assessed it under the same rules as the other claims about cost, without reference to AI. The Committee of Advertising Practice (CAP), which writes the UK code for non-broadcast advertising, including online advertising, advised in June 2026 that an advertiser is responsible for any AI-generated content it uses in an advertisement. In the UK, the Competition and Markets Authority (CMA) can impose up to £300,000 or, if higher, 10% of turnover where a breach of consumer protection law harms consumers collectively. In the US, the Federal Trade Commission (FTC) can seek up to $53,088 per violation from a business that knowingly breaches its rule on fake reviews and testimonials. This piece sets out what a marketing team needs to assess in each variation before it is released, using an illustrative meal-kit campaign.*

## Two ASA rulings on claims in paid advertising

In an April 2026 UK case, the ASA examined two paid Google search listings, seen in July 2025, for an accident claims management company. It identified them for investigation following intelligence from its Active Ad Monitoring system. One listing’s headline included “Guaranteed 100% Free Service”. In its response, the company told the ASA that the claim “100% Free Service” had been generated by AI, that it had since removed it and strengthened its internal approval process, and that future advertising copy would be approved by one of its directors.

The ASA concluded that the listings and their landing page misled consumers about the nature of the service on offer, in part because they did not make clear that the company handled only non-fault claims. It also found that they implied there would be no cost to the consumer, although that was not true in every case, and that the savings claims overstated the potential benefit. The breaches it found were of rules 3.1, 3.3 and 3.9 of the UK Code of Non-broadcast Advertising and Direct and Promotional Marketing (the CAP Code), and of rules 3.17 and 3.38 on the cost and savings claims.

The ruling shows how the ASA applied the rules to short-form advertising. The ASA recognised that the content space in the search advertisements was limited. In its view, a consumer needed to know that the company handled only non-fault claims before deciding to use the service. The point could have been made in a few words, so the listings themselves should have included it. Rule 3.9 of the CAP Code provides that a marketing communication must not mislead by leaving out “significant limitations and qualifications”. The ASA upheld both issues it had raised.

The ruling has three practical implications for a marketing team that produces advertising with AI. First, although the company told the ASA that AI had generated the claim, the ASA judged it by the same rules as the other cost claims, without any reference to AI in its assessment. Second, limited space was no reason to leave out a limitation on the service that could be stated briefly. Third, the ASA raised the challenge itself, drawing on a monitoring system that, in the ASA’s words, “uses AI to proactively search for online ads that might break the rules”. The ASA says the system sorts through large volumes of advertisements and flags likely non-compliant ones for its experts to review, so that they can act more quickly. For a marketing team, a non-compliant AI variation may therefore be more likely to be found, whether or not a consumer or a competitor complains.

Under the ruling, the advertisements must not appear again in the same form. A business that does not comply with an ASA decision can face sanctions. The ASA can ask search engines to remove paid search listings that break the rules, and can refer an advertiser that persistently does so to Trading Standards. CAP can issue Ad Alerts that advise media owners not to provide services to the business concerned, including advertising space. The [earlier piece in this series](https://www.intercepta.ai/blog/brand-reputation-published-rulings/?utm%5Fsource=blog&utm%5Fmedium=internal&utm%5Fcampaign=ai-marketing-brand-control-20261029&utm%5Fcontent=inline) sets out the consequences for a brand once a ruling is published under its name.

The ASA has also ruled on a price offer that misled consumers about what they were buying. In November 2025, in a ruling that made no reference to AI, the ASA upheld complaints about a paid social media advertisement for an online course provider that showed “98% OFF” and a price of £0.99 beside a struck-through price of £49.99\. In the ASA’s assessment, those claims shown together gave the impression of a standalone course offered at a discount as a one-off purchase. The price was in fact available only with a 30-day free trial that renewed automatically into a £269 yearly subscription unless cancelled. CAP’s enforcement notice on free trials and promotional subscription offers, published in March 2023, sets out how offers of this kind must be presented. Under the notice, a reference to “T&Cs apply” is not enough. The advertisement itself must make clear whether the trial becomes a paid subscription by default, and how much the customer commits to paying if it continues. A marketing team therefore assesses each variation for the impression its claims give in combination, and for whether those conditions are stated in it.

![Two ASA rulings. Search listings omitted a non-fault limitation and implied no cost, which was untrue in some cases. The advertiser said AI generated the free-service wording. A £0.99 course offer required a trial renewing at £269 yearly unless cancelled.](https://www.intercepta.ai/blog/content/images/2026/10/fig1-two-asa-rulings--2400x1400--v2.png)

The figure shows findings from the April 2026 ruling on paid search listings and the November 2025 ruling on a paid social media advertisement. Only the April ruling describes a statement by the advertiser that a claim had been generated by AI.

## UK and US regulators on AI-generated advertising

In June 2026, CAP, whose code the ASA enforces, published advice stating that its rules apply to AI-generated content in the same way as to content produced by any other means. CAP adds that responsibility for any AI output used in an advertisement stays with the advertiser, who must always “sense-check” that output. In its view, a celebrity endorsement generated by AI is likely to breach the Code where consumers could take it for a genuine one. The ASA and CAP Annual Report 2025 describes generative AI as making it easier to produce advertisements in volume and to personalise them. It also shows that the number of complaints the ASA assessed about AI-generated advertising content grew in 2025, and that its Active Ad Monitoring system processed nearly 60 million advertisements that year.

Since 6 April 2025, under the Digital Markets, Competition and Consumers Act 2024, the CMA can also impose a penalty directly for an infringement of consumer protection law that harms the collective interests of consumers. The maximum fine is £300,000 or 10% of the business’s total turnover, whichever is higher. The Act treats fake consumer reviews as unfair in all circumstances, so every review or rating a marketing team publishes needs evidence that it is based on a person’s genuine experience.

In the United States, the FTC announced its rule on consumer reviews and testimonials in August 2024, giving AI-generated fake reviews as an example of those falsely attributed to someone who does not exist. Under the rule, a business must not create or sell reviews or testimonials of this kind. Buying the reviews, or disseminating the testimonials, is also prohibited where the business “knew or should have known” that they were fake or false. For each knowing violation of the rule, the FTC can seek a civil penalty of up to $53,088, the level that continues to apply during 2026\. By December 2025, FTC staff had written to 10 unnamed companies to warn them of possible breaches and of those penalties. The FTC stated that the letters were not a formal finding that any recipient had violated the rule, and its release did not say that they concerned AI-generated content. In a US variation, the marketing team should be able to match each review or testimonial to a real person’s own words or experience on record, and remove any that does not match.

The FTC’s Endorsement Guides contain no exception for an endorsement or testimonial drafted with AI, whether in an influencer brief or on a product page. Under the Guides, an endorsement must reflect what the endorser honestly thinks, believes, has found or has experienced. The advertiser can also be liable where such an endorsement conveys a misleading or unsubstantiated claim. For any endorsement drafted with AI, the marketing team should therefore have the endorser confirm that the wording reflects their honest opinion or experience, and remove any claim about the product that the business cannot substantiate.

## One brief, 40 variations

Marketing teams that publish AI-generated content in volume face a question of consumer trust. In May and June 2026, Gartner surveyed 1,006 US consumers. Of those surveyed, 65% said they believe brands produce too much AI-generated content in their marketing, and 57% said this content has made them trust brand messaging less overall.

Brand guidelines can be written into prompts, templates and approved libraries, and those controls govern how content looks and sounds. The regulatory exposure in a piece of content depends on its claims, its audience and the market where it is published. A variation that adds a claim or an offer, or that will run in another market, therefore needs its own assessment, and those variations can only be found by reviewing every one.

In an illustrative example, the marketing team at a meal-kit company has approved copy for an autumn promotion in the United Kingdom and the United States. The copy offers new customers half price on their first box, followed by the weekly price until they cancel. The team asks an AI tool for variations to use across search, paid social media and email, and the tool returns 40\. Thirty-seven restate the approved claims, with their conditions, for the market where each will run. The other three, a share chosen for illustration, change the copy. One shows the offer without its conditions, a second adds “Rated five stars by our customers”, and the third includes “Carbon-neutral delivery”.

Some teams approve a whole batch of variations once a few have been read, on the assumption that the rest will be similar. In the meal-kit example, the reviewer picks four of the 40 at random to compare with the agreed copy. All of them are consistent with it, so the whole set is approved. None of the three variations that change the copy is in the sample, so all three are published unread. The figure below shows how often all three would be left out of a random selection of four, or even ten. Once published, each variation is an advertisement, and the rules apply to it whether anyone reviewed it.

![Grid of 40 illustrative AI variations, three differing from the approved claims and a random sample of four outlined. The chance of none being sampled is 72%, 90% when only one differs, and 41% with ten sampled.](https://www.intercepta.ai/blog/content/images/2026/10/fig2-sampling-40-variations--2400x1400--v3.png)

With only one of the 40 variations differing from what was approved, it would be left out of a random sample of four in nine cases out of ten. The example is illustrative and does not describe the output of any validation tool.

Each of the three variations needs its own assessment before release, against rules that differ between the UK and the US. The figure below shows the half-price offer, the five-star rating and the carbon-neutral claim beside the approved copy, with what each requires in the UK and in the US.

![Table of three illustrative AI variations, a half-price offer without its conditions, a five-star rating and a carbon-neutral claim, setting out what each needs before release and the UK and US rules that may apply.](https://www.intercepta.ai/blog/content/images/2026/10/fig3-three-variations-two-markets--2400x1400--v4.png)

The copy is illustrative, written for this piece without reference to any campaign or tool.

### The half-price offer

The half-price variation shows the offer without the conditions in the approved copy. The discount applies only to a new customer’s first box, and the weekly price is charged after it until they cancel. Like the advertisements in both rulings above, the variation may mislead UK consumers about what is on offer. Under CAP’s enforcement notice on free trials and promotional subscriptions, the variation itself must make clear that the weekly price is charged by default after the first box, and how much the customer commits to paying if they do not cancel. The marketing team therefore restores those conditions in the copy or deletes the discount.

In the United States, an advertisement that leaves out the material conditions of an offer may be deceptive under section 5 of the Federal Trade Commission Act. Subscriptions sold online that continue until cancelled are also subject to the Restore Online Shoppers’ Confidence Act (ROSCA). The Act provides that the seller must set out every material term of the transaction in text, clearly and conspicuously, before it takes any billing information. The customer’s express informed consent is also needed before charging, as is a simple means of stopping the recurring payments. The FTC enforces the Act in the same way as its own rules. The marketing team either removes the half-price offer or states its conditions in the variation itself.

### The five-star rating

The second variation adds “Rated five stars by our customers”, which depends on genuine ratings, presented in a way that does not mislead. In the United Kingdom, rule 3.44 of the CAP Code provides that a marketing communication must not include fake consumer reviews. Under rule 3.46, publishing reviews in a misleading way includes withholding negative reviews while publishing positive ones. The rule applies equally to giving positive reviews more prominence, or to presenting a rating in a way that misleads. The CMA can also impose the penalty set out above for fake reviews and for misleadingly presented reviews or ratings, where the practice harms consumers collectively. As rule 3.7 requires, the marketing team holds documentary evidence before release that the ratings behind the five-star claim are genuine and that the claim reflects them accurately. Without that evidence, it removes the claim from the variation.

In the United States, that five-star wording may in its own right be deceptive under section 5\. The FTC’s 2024 rule on reviews and testimonials also governs any customer ratings behind the claim that were submitted to a website or platform that displays reviews. Under that rule, a business must not create reviews that appear to come from people who do not exist or have not used the product. On a website or platform it operates, it also may not hide reviews because they are negative while presenting the remaining ones as the full set or the majority of those it has received. For a US variation, the marketing team needs the same evidence, or withdraws the claim.

### The carbon-neutral claim

“Carbon-neutral delivery” is added in the third variation. In the United Kingdom, rule 11.1 of the CAP Code provides that “the basis of environmental claims must be clear”. In its guidance on carbon-neutral and net zero claims, CAP advises advertisers to say whether, and how far, they are reducing their emissions or relying on offsetting. For the marketing team, that means setting this out in the variation, with the supporting evidence already on file.

In the United States, the FTC Green Guides apply to an environmental claim such as carbon-neutral delivery. Under section 260.2, each reasonable interpretation of the claim must be truthful and not misleading. Consumers could, for example, read the claim as covering every delivery, or as meaning that emissions were reduced without relying on offsetting. The marketer also needs a reasonable basis for each interpretation before publication. The marketing team holds that substantiation ahead of US release.

### Superlatives and false time limits

In some drafts, an AI tool might introduce a superlative, or a comparison with a rival product, that the team cannot substantiate. It can also insert a countdown or closing date that the offer does not have. The wording after either edit can differ little from the approved version, yet the advertisement may no longer be compliant.

For the ASA in the UK, an unqualified superlative amounts to a comparison with every competing product. Unless the claim is obvious puffery, the marketer needs evidence that covers each of those products. Rule 3.30 of the CAP Code provides that a marketing communication must not claim a non-existent time limit for a product or its terms, where the aim is “to deprive consumers of the time or opportunity to make an informed choice”. The marketing team keeps a superlative or comparison that needs substantiation only where it holds the evidence before release. It also deletes any countdown or closing date that it could not publish without breaching rule 3.30\. In the United States, either edit may be deceptive under section 5, so the marketing team assesses each US variation against that standard.

### Labelling AI content

Labelling AI content is a separate question from whether its claims are accurate. CAP stated in May 2025 that UK law contains no blanket requirement to disclose the use of AI in advertising. In its view, the first question to ask is whether leaving out a disclosure of AI use would be likely to mislead the audience. For each UK variation, the marketing team therefore makes that judgement. CAP added that, where the substance of an advertisement misleads, disclosing the use of AI is very unlikely on its own to mitigate the harm. Whatever the team decides on disclosure, it still validates the claims in the variation.

In the United States, three states have so far passed laws that provide for a disclosure in commercial advertising in defined cases. In New York, a synthetic performer used in an advertisement must be disclosed conspicuously within it. The requirement covers a business that produces the advertisement for goods or services it offers, where the business has actual knowledge that the performer is there. This duty, in subdivision 3 of section 396-b of the General Business Law, has been in force since 9 June 2026\. It may also apply to an agency that produces the advertisement. A synthetic performer here means a figure made or altered with AI or software to appear as a human performer in images or video, who cannot be recognised as any real, identifiable performer. The law does not cover audio-only advertisements. The marketing team should add the disclosure to any New York variation it knows contains a synthetic performer, and agree with its agency which of them adds it.

In California, SB 1050 applies from 1 January 2027\. A person who creates an advertisement and causes it to be published must give a clear and conspicuous disclosure of any synthetic performer that the advertisement prominently includes. Unlike New York’s, California’s definition also covers a voice made wholly or partly with generative AI, where it gives the realistic impression of a human performer who cannot be recognised as any real, identifiable person. Voices in a California variation therefore need the same assessment as figures, and the marketing team should add the disclosure for any synthetic performer the advertisement prominently includes.

Pennsylvania’s law takes effect on 21 December 2026\. An advertiser that uses content generated or significantly modified by AI, other than text alone, must give a clear and conspicuous disclosure where it uses that content in a way that creates a false consumer perception about the nature, origin or authenticity of the goods or services advertised. A Pennsylvania variation should include a clear and conspicuous disclosure if its AI-made or AI-altered visuals or audio could give a false impression of what is advertised.

## Platforms, agencies, markets and sectors

Advertising platforms can assemble an advertisement from several assets. Google allows an advertiser to supply up to 15 headlines and 4 descriptions for a single responsive search advertisement, and it selects among them to show different combinations in different orders. A headline can therefore appear without the description that contains its conditions, in a combination that no one in the marketing team reviewed as a whole. In the UK, CAP’s June 2026 advice states that an advertiser that uses an automated AI platform to generate or distribute its advertisements is still responsible for them.

![A meal-kit search advertisement built from three headlines and two descriptions. Without a pin, the half-price headline can appear with a recipes description and no conditions. With the conditions description pinned to description position 1, they are shown every time.](https://www.intercepta.ai/blog/content/images/2026/10/fig4-search-ad-combinations--2400x1400--v2.png)

Without a pin, the platform can show the half-price headline with a description that does not state the offer’s conditions. In the third column, the description that sets out those conditions is pinned to position 1.

Under rule 1.8 of the UK’s CAP Code, the advertiser, as the marketer, bears primary responsibility for the compliance of content that agencies produce on its behalf. Those agencies, like other service suppliers, also commit to following the Code. In a 2024 survey by the World Federation of Advertisers, 80% of the multinational brand owners taking part had expressed concerns about how their creative and media agencies use generative AI on their behalf. There were 54 responses, from 48 companies with combined annual marketing spend of $102 billion.

A marketing team can apply three working controls. First, a claim that depends on conditions is supplied to a platform only in an asset that sets them out. In a responsive search advertisement, where a headline has no room for those conditions, the description containing them is pinned to description position 1\. Google’s guidance is that a description pinned there appears every time the advertisement is shown. Pinning can affect the Ad Strength rating Google gives the advertisement, but the rating has no bearing on which conditions the advertisement must disclose. Second, agencies are asked to return each variation with its intended market named and its claims set against the approved claim. Third, a claim approved for one market is assessed again before it is published in another, since the rules, and the evidence they require, differ between markets.

In the release cycle, the assessment takes place at the hand-off from the agency or the in-house team to the platform. Each asset, including every headline and email, is assessed before it is uploaded, so that any issue is remediated in the copy within the same cycle. Only assets assessed for their market are then supplied to the platform.

In regulated sectors, each variation is also subject to the sector’s own rules. For consumer credit, rule CONC 3.5.3R of the Financial Conduct Authority (FCA, UK) applies to a financial promotion that shows an interest rate or a figure for the cost of the credit, including one generated by AI. Subject to limited exceptions, such a promotion must include a representative example. CAP’s advice on financial products indicates that the ASA is unlikely to cover advertising for the technical aspects of products from FCA-regulated businesses and likely to cover their non-technical aspects. A single variation from a lender or insurer may therefore need to be assessed against both the FCA rules for its sector and the CAP Code.

In the United States, FTC staff guidance states that, as a general rule, claims about the health benefits or safety of health-related products “require substantiation in the form of competent and reliable scientific evidence”. For any health benefit claim that an AI tool shortens, the marketing team holds that evidence for the new wording.

## Review at the volume of AI output

Survey findings indicate that full review of AI output is not yet usual, although neither survey cited here covers the review of marketing content specifically. McKinsey surveyed respondents across regions, industries and functions in July 2024 and published the results in March 2025\. Among respondents whose organisations use generative AI, 27% said that employees review all of the AI’s output before use, and a similar share said that 20% or less is reviewed. The meal-kit reviewer reads four of the 40, or 10%, so the team’s practice matches what this second group reported. In The CMO Survey for 2026, US marketing leaders expected their companies’ use of AI or machine learning in optimising and automating marketing to rise, on average, from 24% of the time to 56% within three years.

In the meal-kit campaign, reviewing the full volume of output means validating the claims in each of the 40 against the rules of its market, since a sample of four would leave 36 unreviewed. Reviewers begin with any potential issues flagged for each variation, although a variation without a flag still needs approval. Some matters need their own review, including the evidence that substantiates each claim and any imagery an AI tool produces. In the UK, CAP’s June 2026 advice states that advertisers should “sense-check” AI outputs to ensure that images are socially responsible and free from harmful, discriminatory stereotypes. Each decision is recorded before release, so the advertiser can show what was assessed, against which rules, the evidence relied on, the result and who approved it.

In its response to the ASA’s challenge, the company in the April 2026 UK ruling said that future advertising copy would be approved by one of its directors. A marketing team that approves each of the 40 variations before publication can remediate the same kind of issue while the copy is still a draft.

![The same 40-variation grid with every variation validated before release, the three differing variations annotated with rule and remediation, and an example record entry listing variation, rule, issue, revision, evidence, approver and date.](https://www.intercepta.ai/blog/content/images/2026/10/fig5-every-variation-validated--2400x1400--v5.png)

The figure shows the review each variation needs. It does not reproduce the output of Intercepta AI or any other tool, and it does not show which claims any tool would flag.

## Validate before you publish. Keep the record.

An AI tool can produce more variations from one brief than a sample review covers, and a variation that changes an approved claim can easily be left out of one. In the April 2026 UK ruling, the ASA made no distinction between the cost claim that the advertiser attributed to AI and the other cost claims. In the UK and the US, the rules apply to each variation an advertiser publishes, as they do to any other advertisement. Intercepta AI was built for marketing teams that produce content with AI at that volume, and for the compliance teams that approve it.

The platform validates the claims in text and on websites, and on paid plans in audio and video, against 1,100+ regulatory rules across the United States, United Kingdom, European Union, Canada, Australia and New Zealand. Each scan returns a documented report that maps each issue to the specific regulation it references, with remediation guidance for your compliance team to review. For a responsive search advertisement, each headline and description is validated as text before upload. The platform does not assess the combinations that Google assembles from them, so the marketing team decides which description to pin. Batch audit of a full set of variations is available on agency and enterprise plans. Marketing teams validate before publication, compliance teams hold a documented record of each validation, and every decision on every finding remains in the remit of your compliance team.

[Scan the copy in one of your AI variations →](https://www.intercepta.ai/?utm%5Fsource=blog&utm%5Fmedium=cta&utm%5Fcampaign=ai-marketing-brand-control-20261029&utm%5Fcontent=cta-button)

## Cross-references

- For how regulators and courts make the outcome of a case against an advertiser public, see [Advertising rulings and brand reputation: when the advertiser is named, and the notices it may have to issue itself](https://www.intercepta.ai/blog/brand-reputation-published-rulings/?utm%5Fsource=blog&utm%5Fmedium=internal&utm%5Fcampaign=ai-marketing-brand-control-20261029&utm%5Fcontent=cross-reference).
- For the European Union’s AI transparency obligations and their dates, see [EU marketing compliance in 2026: what applies and when](https://www.intercepta.ai/blog/eu-marketing-compliance-2026/?utm%5Fsource=blog&utm%5Fmedium=internal&utm%5Fcampaign=ai-marketing-brand-control-20261029&utm%5Fcontent=cross-reference).

## Sources

### United Kingdom

- [Committee of Advertising Practice, AI and deepfakes: four things advertisers need to know before they hit “run”, CAP News, 11 June 2026.](https://www.asa.org.uk/news/ai-and-deepfakes-four-things-advertisers-need-to-know-before-they-hit-run.html?ref=intercepta.ai)
- [Committee of Advertising Practice, Disclosure of AI in advertising: striking the balance between creativity and responsibility, CAP News, 29 May 2025.](https://www.asa.org.uk/news/disclosure-of-ai-in-advertising-striking-the-balance-between-creativity-and-responsibility.html?ref=intercepta.ai)
- Advertising Standards Authority, ruling A25-1299398, 29 April 2026\. Subject firm anonymised in this piece under house policy; party-naming URL withheld under the sources-link policy.
- Advertising Standards Authority, ruling G25-1309531, 26 November 2025\. Subject firm anonymised in this piece under house policy; party-naming URL withheld under the sources-link policy.
- [Committee of Advertising Practice, CAP Code (Edition 12), rule 1.8 in section 1 (Compliance).](https://www.asa.org.uk/type/non%5Fbroadcast/code%5Fsection/01.html?ref=intercepta.ai)
- [Committee of Advertising Practice, CAP Code (Edition 12), rules 3.1, 3.3, 3.7, 3.9, 3.10, 3.17, 3.30, 3.38, 3.44 and 3.46 and the principle on unqualified superlative claims in section 3 (Misleading advertising).](https://www.asa.org.uk/type/non%5Fbroadcast/code%5Fsection/03.html?ref=intercepta.ai)
- [Committee of Advertising Practice, CAP Code (Edition 12), rule 8.17 in section 8 (Promotional marketing).](https://www.asa.org.uk/type/non%5Fbroadcast/code%5Fsection/08.html?ref=intercepta.ai)
- [Committee of Advertising Practice, CAP Code (Edition 12), rules 11.1 and 11.3 in section 11 (Environmental claims).](https://www.asa.org.uk/type/non%5Fbroadcast/code%5Fsection/11.html?ref=intercepta.ai)
- [Committee of Advertising Practice, Updated environment guidance: carbon neutral and net zero claims in advertising, 10 February 2023.](https://www.asa.org.uk/news/updated-environment-guidance-carbon-neutral-and-net-zero-claims-in-advertising.html?ref=intercepta.ai)
- [Advertising Standards Authority and Committee of Advertising Practice, Sanctions.](https://www.asa.org.uk/codes-and-rulings/sanctions.html?ref=intercepta.ai)
- [Committee of Advertising Practice, Advertising of free trials and promotional offer subscription services, enforcement notice, 30 March 2023.](https://www.asa.org.uk/resource/enforcement-notice-advertising-of-free-trials-and-promotional-offer-subscription-services.html?ref=intercepta.ai)
- [Committee of Advertising Practice, Financial products and services: general, Advice Online, 25 March 2026.](https://asa.org.uk/advice-online/financial-products-and-services-general.html?ref=intercepta.ai)
- [Advertising Standards Authority, Our Active Ad Monitoring system, 14 July 2023.](https://www.asa.org.uk/news/our-active-ad-monitoring-system.html?ref=intercepta.ai)
- [Advertising Standards Authority and Committee of Advertising Practice, ASA and CAP Annual Report 2025, 14 April 2026.](https://www.asa.org.uk/resource/asa-and-cap-annual-report-2025.html?ref=intercepta.ai)
- [Financial Conduct Authority, FCA Handbook, Consumer Credit sourcebook, CONC 3.5.3R.](https://handbook.fca.org.uk/handbook/conc3/conc3s5?ref=intercepta.ai)
- [Digital Markets, Competition and Consumers Act 2024, sections 148(1) and 182(6) and Schedule 20, paragraph 13, in force 6 April 2025 (SI 2025/272).](https://www.legislation.gov.uk/ukpga/2024/13/contents?ref=intercepta.ai)

### United States

- [Federal Trade Commission Act, section 5, 15 U.S.C. § 45(a)(1).](https://www.law.cornell.edu/uscode/text/15/45?ref=intercepta.ai)
- [Restore Online Shoppers’ Confidence Act, 15 U.S.C. §§ 8403(1) to (3) and 8404(a).](https://www.law.cornell.edu/uscode/text/15/chapter-110?ref=intercepta.ai)
- [Federal Trade Commission, “Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials”, press release, 14 August 2024.](https://www.ftc.gov/news-events/news/press-releases/2024/08/federal-trade-commission-announces-final-rule-banning-fake-reviews-testimonials?ref=intercepta.ai)
- [Trade Regulation Rule on the Use of Consumer Reviews and Testimonials, 16 C.F.R. Part 465, sections 465.1(d), 465.2 and 465.7(b).](https://www.ecfr.gov/current/title-16/chapter-I/subchapter-D/part-465?ref=intercepta.ai)
- [Federal Trade Commission, “FTC Warns 10 Companies About Possible Violations of the Agency’s New Consumer Review Rule”, press release, 22 December 2025.](https://www.ftc.gov/news-events/news/press-releases/2025/12/ftc-warns-10-companies-about-possible-violations-agencys-new-consumer-review-rule?ref=intercepta.ai)
- [Federal Trade Commission, Civil Penalty Inflation Adjustments, 91 FR 58446, FR Doc. 2026-18853, 15 September 2026.](https://www.federalregister.gov/d/2026-18853?ref=intercepta.ai)
- [Federal Trade Commission, Adjustment of civil monetary penalty amounts, 16 C.F.R. § 1.98(d) (section 5(m)(1)(A) of the FTC Act).](https://www.ecfr.gov/current/title-16/chapter-I/subchapter-A/part-1/subpart-L/section-1.98?ref=intercepta.ai)
- [Federal Trade Commission, Guides Concerning Use of Endorsements and Testimonials in Advertising, 16 C.F.R. Part 255, sections 255.0(b) and 255.1(a) and (d).](https://www.ecfr.gov/current/title-16/chapter-I/subchapter-B/part-255?ref=intercepta.ai)
- [Federal Trade Commission, Guides for the Use of Environmental Marketing Claims, 16 C.F.R. Part 260, sections 260.2, 260.3(b) and 260.5.](https://www.ecfr.gov/current/title-16/chapter-I/subchapter-B/part-260?ref=intercepta.ai)
- [Federal Trade Commission, Health Products Compliance Guidance, December 2022, section II.A.](https://www.ftc.gov/business-guidance/resources/health-products-compliance-guidance?ref=intercepta.ai)
- [New York General Business Law, section 396-b, added by S8420A (Chapter 617 of the Laws of 2025), in effect from 9 June 2026.](https://www.nysenate.gov/legislation/laws/GBS/396-B?ref=intercepta.ai)
- [California Business and Professions Code, section 17610, added by SB 1050 (Chapter 246, Statutes of 2026), in effect from 1 January 2027.](https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill%5Fid=202520260SB1050&ref=intercepta.ai)
- [Pennsylvania Artificial Intelligence Deceptive Advertising Act, Act No. 99 of 2026 (Senate Bill 806, printer’s number 1808), sections 3, 4 and 6, approved 22 October 2026, in effect from 21 December 2026.](https://www.palegis.us/legislation/bills/2025/sb806?ref=intercepta.ai)

### Platform documentation

- [Google Ads Help, About responsive search ads.](https://support.google.com/google-ads/answer/7684791?ref=intercepta.ai)

### Research and surveys

- [McKinsey and Company, The state of AI: How organizations are rewiring to capture value, 12 March 2025.](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-how-organizations-are-rewiring-to-capture-value?ref=intercepta.ai)
- [The CMO Survey (Duke University Fuqua School of Business, Deloitte and the American Marketing Association), The CMO Survey Topline Report 2026, page 11.](https://cmosurvey.org/wp-content/uploads/2026/03/The%5FCMO%5FSurvey-Topline%5FReport-2026.pdf?ref=intercepta.ai)
- [Gartner, “Gartner Marketing Survey Finds 35% of Consumers Rely on Influencers Less Due to AI”, press release, 22 September 2026.](https://www.gartner.com/en/newsroom/press-releases/2026-09-22-gartner-marketing-survuey-finds-35-percent-of-consumers-rely-on-influencers-less-due-to-ai?ref=intercepta.ai)
- [World Federation of Advertisers, Eighty percent of brands have concerns about agency use of GenAI, 17 September 2024.](https://wfanet.org/knowledge/item/2024/09/17/eighty-percent-of-brands-have-concerns-about-agency-use-of-genai?ref=intercepta.ai)