On 11 August 2026, OpenAI launched advertisements on ChatGPT in the UK, enabling retailers to now advertise directly to ChatGPT users. Whilst ChatGPT may not be the first AI platform to incorporate advertising, the development is a timely prompt for advertisers to consider the integration of advertisements into conversational AI services as an emerging channel to reach consumers. 

The potential is significant, rather than targeting someone simply because they searched for 'running shoes', an advertiser could engage a user who is discussing how to train for their first 10K race – reaching them at a moment of genuine intent and relevance.

So, what should retailers know before considering this channel?

How do ChatGPT ads work?

Ads appear below ChatGPT's responses for users on Free and Go plans (the more expensive Plus and Pro tiers are not included), they are labelled as 'Sponsored' and visually separated from the answer. OpenAI, the owner of ChatGPT, says that ads do not influence the chatbot responses and that individual conversations are not shared with advertisers. Instead, in matching ads with consumers, OpenAI has stated that it relies on signals including the context and intent of the current conversation which will be matched with advertisers’ descriptions of the type of conversations in which their products may be relevant. Where personalised ads are enabled, previous conversations can be used as signals.

What advertising rules apply?

The usual UK advertising rules still apply. The UK Code of Non-broadcast Advertising and Direct & Promotional Marketing (CAP Code) requires that the commercial intent is made clear. This reflects the Digital Markets, Competition and Consumers Act 2024 (DMCCA), which prohibits misleading omissions that include failing to identify the commercial intent of a commercial practice (including communications) and also bans editorial content from being used to promote a product without making it clear that the promotion has been paid for. 

The CAP Code requires marketing communications to be obviously identifiable as advertisements, and their commercial intent must be clear where it is not otherwise apparent. Advertisement features need to be labelled as such. The key issue is going to be whether the regulators think that the average consumer will understand that an ad is an ad. This requirement will be particularly important in conversational AI interfaces, where users may view AI generated responses as presenting information that has been independently gathered, or assuming that content is provided by a machine on a neutral basis. 

OpenAI’s current design seeks to separate paid for content from the answer generated and labels it as 'Sponsored'. However, this may not be sufficient. The Advertising Standards Authority has previously taken action against advertisements on social media, especially by influencers because the labels used were not considered inadequate. Guidance from the Committee of Advertising Practice specifically warns companies against using 'Sponsored' as a label, on the basis it is too ambiguous and open to interpretation. 

The usual rules on misleading advertising also continue to apply. Claims must be accurate and substantiated, while existing requirements relating to pricing, promotions, environmental claims, health and beauty claims and advertising to children remain applicable.

Therefore, all advertisers, including retailers should be aware that whilst the AI company controls how and where an advertisement is displayed and bears its own responsibility for maintaining a distinction between output that is and isn’t advertising, retailers will ultimately be held responsible for ensuring that advertising is compliant with the DMCCA and CAP Code. This is reinforced  by OpenAI’s own Ad Policy which requires advertisers to ensure that ads are clearly distinguishable from ChatGPT responses, and states that OpenAI may reject advertising content that imitates ChatGPT in a way that could lead users to believe that the ad forms part of the ChatGPT response.

What about the UK GDPR and PECR?

OpenAI is also subject to data protection obligations because information from conversations, including personal data, is used to determine if advertisements should be shown and what form they will take.

OpenAI states that ads will not appear alongside sensitive or regulated topics, including personal health, mental health or politics. Its Advertising policy also excludes a range of “sensitive user contexts” from ads targeting.

This is an important safeguard, but it raises an interesting question – will determining if a conversation is sensitive itself involve processing or inferring sensitive personal data? The Information Commissioner’s Office (ICO) states that storage and access technologies used for online advertising, including ad measurement, require consent under PECR; any associated processing of personal data must also comply with the UK GDPR. OpenAI’s Privacy Policy identifies legitimate interests as the relevant lawful basis for showing users generic ads and consent as the lawful basis for showing personalised ads. The published Privacy Policy does not separately identify an Article 9 condition for special category data processing in connection with advertising. 

OpenAI has stated that it will not share conversations with advertisers, arguably reducing retailers’ direct exposure to this upstream processing. For retailers, the more significant privacy issue may arise after a user clicks on an advertisement. OpenAI offers conversion measurement through a Pixel and Conversions API and therefore still activates the consent requirements of PECR. Retailers should therefore review their cookie consent mechanisms, privacy notices and adtech documentation before implementing OpenAI’s measurement tools.

What else should retailers be aware of?

The Age Appropriate Design Code (Children’s Code) is relevant to services likely to be accessed by children. OpenAI states that it does not show advertisements to accounts that it determines belong to people under the age of 18, using account information and age prediction to support this measure. Retailers should nonetheless be aware that age prediction is an imperfect tool, and that the ICO expects services to take a risk-based approach to child protection – meaning that reliance on OpenAI's measures alone may not be sufficient to discharge a retailer's own obligations under the Children's Code where their products or services are capable of being accessed by or are directed at children.

Finally, sector-specific rules continue to apply. Financial promotions, for example, remain subject to the requirements of the Financial Services and Markets Act (FSMA) and the Financial Conduct Authority (FCA), although OpenAI currently generally prohibits advertisements for financial services outside the United States. 

For retailers, advertising within conversational AI represents a genuinely new route to consumers, one that offers contextual relevance and proximity to purchasing intent in a way that traditional search advertising does not. The channel is novel; the legal and regulatory framework governing it is not. Retailers that move quickly to exploit this opportunity without first ensuring that their advertising is clearly identified, accurately presented, and responsibly targeted risk enforcement action. Retailers should treat entry into this channel as a prompt for a broader review of advertising compliance, rather than simply an extension of existing digital marketing activity.