UK consumer protection rules require paid-for editorial content to be disclosed to consumers. As AI assistants increasingly mediate access to information, Boris Dzhingarov argues that these disclosures should also be machine-readable, so that the commercial context of paid-for content is not lost when it is processed and presented by AI systems.
As consumers increasingly access information through AI systems, existing disclosure requirements may no longer ensure that the commercial context of paid-for content reaches them. The Competition and Markets Authority (CMA) should therefore consider treating machine-readable disclosure as part of compliance, requiring paid-for content to carry a standardised indication in its page markup and structured data alongside the existing visual disclosure. This would preserve an existing consumer protection principle (that people should be able to identify paid promotions) as the way they access information changes.
The disclosure gap created by AI
Ask an AI assistant about a company, product or service, and its response may draw on information from articles that were originally paid for by the company. This raises an important question for consumer protection. Can people tell when information has been influenced by advertising if they encounter it through an AI-generated answer rather than on the original website?
UK law already requires paid-for content to be clearly identified. Under the Digital Markets, Competition and Consumers Act 2024, in force since April 2025, businesses cannot pay to promote a product through editorial content without making the commercial relationship clear to consumers.
These rules were designed largely for a world in which people read the original content. But the growing use of AI assistants is changing how information reaches consumers.
An AI system may draw on the text of an article without carrying across the label that identifies it as paid-for content. The information can therefore reach the consumer without the commercial context that accompanied it.
Why existing disclosures may not reach consumers
Paid-for articles are generally labelled so that readers can identify their commercial nature. The disclosure almost always lives in what a designer would call the ‘furniture’ of the page: it might appear above the headline, alongside the author’s names or at the bottom of the article.
All of these placements would satisfy a statutory test. For somebody reading the original webpage, they indicate that payment was involved. But the disclosure is often separate from the main body of text in the article itself.
This becomes important when an AI assistant uses information from the page.
An assistant may draw on material from several sources and bring it together into one answer. Information contained in a paid-for article could therefore appear in that answer, without the original disclosure from the article’s ‘furniture’ appearing alongside it.
Even if the AI assistant provides a link to its source, a consumer may read confident prose and rely on the perceived accuracy of the answer without visiting the original webpage. This creates a potential gap between the information that someone receives through an AI assistant and the disclosure that they would have seen when reading the original article.
That does not necessarily mean that anyone has broken the law.
The advertiser may have paid for clearly labelled content, the publisher may have labelled the article appropriately and the AI assistant may have processed information available on a public website, as it is allowed to do. The wider question is whether disclosure rules designed for one way of accessing information remain effective as people’s habits change.
Why this matters for consumer protection
This would be a straightforward consumer protection issue if the questions people asked AI assistants were narrow. They are not. People increasingly ask these systems which energy tariff to choose, whether a care home has a good reputation, or whether a company is trustworthy enough to hold a deposit. These are markets in which regulators have long recognised the need for transparency about paid promotions.
The CMA has spent years addressing hidden advertising on social media, including through detailed principles for platforms and guidance for those posting incentivised content. The Advertising Standards Authority has also been updating its codes to reflect the Digital Markets, Competition and Consumers Act 2024. Much of this assumes that disclosures will reach the person consuming the content. That assumption is becoming less reliable as consumers increasingly obtain information through AI systems.
Two changes could address much of this problem without requiring new primary legislation.
A machine-readable disclosure standard
First, machine-readable disclosures should be treated as part of compliance rather than an optional technical measure. A paid-for article should carry a flag in its page markup and structured data, identifying that the content was paid for. The visual label would remain for human readers, while the machine-readable disclosure would travel with the content when it is ingested by other systems.
Publishers already generate structured data describing, among other things, authorship, publication dates and article types. Adding a field identifying paid-for content would therefore be a relatively modest technical change. The CMA could set out this expectation through its unfair commercial practices guidance, rather than waiting for new legislation.
Extending responsibility through the advertising supply chain
Second, some responsibility should sit with intermediaries in the advertising supply chain. Current enforcement tends to focus on the trader that paid for the promotion and, to a lesser extent, the publisher that carried it. In practice, agencies and networks may have the clearest records of the commercial relationship. Publishers may not know that a freelance contributor has received payment from a third party, particularly where content passes through several intermediaries.
Indeed, I have seen reputable outlets publish content without understanding that a placement fee was paid. Agencies and networks are therefore well placed to ensure that the relevant disclosure is passed through the supply chain. At present, intermediaries seem to be the link in the chain to which these requirements currently apply least directly.
There is an argument that publishers could adopt this approach voluntarily, and some may choose to do so. The difficulty is that unilateral adoption can create a competitive disadvantage. A publisher that consistently identifies paid-for content in machine-readable form, while its competitors do not, may make its content less attractive to the AI systems through which advertisers increasingly seek to reach consumers.
The publisher would therefore bear a cost for greater transparency without necessarily receiving a corresponding benefit. This is a familiar reason for establishing a common regulatory standard rather than relying solely on voluntary compliance.
Applying the approach across borders
There is also a question of jurisdiction. For example, I am based in Bulgaria and place articles on sites registered all over the world. As content may be placed on websites registered in other countries, what can a British regulator do about agencies operating outside the UK?
The answer is that the CMA may still have significant leverage where the relevant trader is subject to UK law. The Digital Markets, Competition and Consumers Act 2024 applies to the trader who commissions the paid-for promotion. Often, this involves a UK trader marketing products or services to UK consumers. Where the legal obligation applies to that trader, a requirement for machine-readable disclosure could be incorporated into its contractual arrangements with agencies and publishers. The obligation could therefore travel through the supply chain in much the same way as existing advertising and compliance requirements.
A longer-term role for AI providers
A further step would be to require AI assistant providers to preserve and, where relevant, surface these disclosures in their answers. That would provide the strongest protection for consumers, but it raises more difficult questions, including how paid-for content should be defined across jurisdictions and how disclosures should be handled when an answer draws on multiple sources. Those issues should be considered, but they do not need to be resolved before improving the underlying disclosure standard.
The more immediate issue is simpler. The wider consumer protection framework was developed when consumers were generally expected to encounter commercial content directly. That assumption no longer holds consistently. A disclosure that is visible on a webpage but lost when the underlying content is processed by an AI system may satisfy the requirements applicable to the original publication while failing to provide meaningful transparency to the consumer who ultimately relies on the information.
Machine-readable disclosure would not require a new model of advertising regulation. It would adapt an existing principle (that consumers should be able to identify paid promotion) to a distribution environment in which information is increasingly consumed indirectly. The objective should be to ensure that the disclosure remains attached to the content, regardless of whether the eventual reader encounters the original page or an AI-generated answer based on it.
About the author: Boris Dzhingarov is the Founder and Chief Executive of ESBO Ltd. He writes about marketing and technology topics for Forbes Agency Council, Entrepreneur and Fast Company Executive Board.
Declaration of interests: The author runs an agency that arranges paid placements in online publications, working with a network of more than 10,000 publishers across approximately a dozen languages. Placement fees range from around $50 on smaller niche sites to five figures on larger publications, with most client purchases falling between approximately $200 and $1,000 per article. The policy proposal set out in this article would apply directly to businesses such as the author's. He therefore has a direct commercial interest in the regulatory approach under discussion.
All articles posted on this blog give the views of the author(s), and not the position of the IPR, nor of the University of Bath.
Respond
