Businesses in ChatGPT: what a European business can claim, and what it signs away

What you can and cannot do from Europe

OpenAI launched Businesses in ChatGPT in beta on 10 November 2026: a space called Business Hub where a business claims its ChatGPT listing, corrects its details, adds a catalogue with prices and availability, and reads usage figures. OpenAI's announcement opens it to “any eligible business located where ChatGPT is available”, which includes the whole of the EU. The help centre describes a narrower scope than the announcement:

The sentence that matters for a baker, hairdresser or restaurant owner is in the help centre, not the announcement: “Local stores and restaurants can currently claim only an existing U.S. listing,” says the help centre. OpenAI gives no date for other countries.

Claiming a listing: the steps and the documents

Business Hub opens from the ChatGPT sidebar once you are signed in. The help centre sets out two successive checks.

1. Prove your link with the business. Depending on what is offered for your listing: a code by text or automated call to the business phone, a code sent to the business email address, a TXT record added to your domain's DNS, or a connected Shopify store. OpenAI then reviews the claim and replies by email.
2. Verify your identity with Persona, OpenAI's verification provider. Without it you cannot publish changes, connect a product feed, see insights or create an agent. OpenAI warns that you may need a government-issued ID, business registration information and proof that you are authorised to represent the business, and that the name, address and any registration or tax number must match official records. The blue verified badge appears only after this step.

Two things to know before you start. A published listing cannot be removed: only drafts can be deleted, and OpenAI Support “is unable to delete, unpublish, or hide a published profile”, according to the help centre. And claiming does not move you up in answers: it “doesn't guarantee increased visibility”, and advertising “doesn't influence ChatGPT's independent answers”, the same page says.

What the insights actually measure

This is the useful part for a small business wondering whether ChatGPT talks about it. Insights cover the past 28 days, according to the help centre:

• Mentions. For an online business, a mention counts when ChatGPT references the business's website; the brand name alone does not count. If ChatGPT recommends you without linking to your site, it will not show up.
• Website clicks. How often people click through from ChatGPT to your site.
• Discovery topics. Common themes in ChatGPT answers that cited you, grouped: neither individual prompts nor conversations.

Figures only appear above reporting thresholds, and a blank metric “doesn't necessarily mean zero activity”, the help centre warns. A small business should expect blanks in the first weeks.

Three clauses to read before you connect a product feed

The Businesses in ChatGPT Terms, last updated on 9 November 2026, apply as soon as you enable, access or use the service. Three passages deserve a careful read by any business that publishes its own content online.

The text and data mining waiver (Section 3). If you provide a product feed, you must let OpenAI's crawlers and agents access your websites, and you and your affiliates “do not exercise or reserve any rights under any text and data mining exception, opt-out, or similar mechanism” with respect to OpenAI, for any of your websites. If such a reservation exists, you agree it “will have no effect” towards OpenAI. In EU law, that reservation is exactly what lets a rightholder refuse the mining of its content: Article 4(3) of Directive 2019/790 makes the commercial mining exception apply only where use “has not been expressly reserved by their rightholders in an appropriate manner, such as machine-readable means in the case of content made publicly available online”. If your robots.txt blocks OpenAI's crawlers to protect your text or photos, connecting a product feed means giving up that block towards OpenAI. The same mechanism sits behind the CJEU case on chatbots and press publishers.

A perpetual licence on basic details (Section 4). Under the terms, the business name, address, contact details, category, opening hours and website you submit for public display are licensed to OpenAI on a “perpetual and irrevocable” basis. The same section states that OpenAI may create and display a listing for your business whether or not you claim it, and supplement it from other sources. It adds that by using Business Hub, you agree that “OpenAI's AI assistant may make calls to you on behalf of our users”, with an opt-out.

The covenant not to sue (Section 10). Under the terms, you “irrevocably and perpetually covenant and agree not to assert or commence” any “Covered Claim” against OpenAI, for yourself and your affiliates. The definition covers claims about the data you provide, including its use for developing and improving OpenAI's products, and about any content, data or materials published on your websites, other than a claim that OpenAI breached the terms. Copyright is not named, but the definition reaches everything online on your sites, before or during the agreement.

For a spare-parts retailer these clauses weigh little. For a photographer, a publisher, a course creator or any business whose website is the product, they deserve a lawyer's read before enabling anything. Whether such clauses hold under a given member state's law is a question the terms do not address.

Google's terms draw a similar line for developers: its new Gemini Decisions API is free to try but must be paid for once European users are served.

What stays in the US: the sponsorable agent

A Business Agent answers customer questions from the business's information, in a separate, labelled conversation. A business can sponsor it and use it as the destination for ChatGPT Ads, according to OpenAI's announcement. It cannot complete purchases, take bookings or check an order: it links to your website, says the help centre. If sponsored agents reach Europe, the EU's proposed Digital Fairness Act would require paid suggestions in shopping agents to be disclosed.

The Business Agents Policy already bans false urgency, pushing a sale after someone declines and collecting sensitive data, and requires the agent to be clear that it is AI. If agents reach Europe, they will meet EU rules on personal data, which the guide to objecting to AI training on your data sets out from the user's side.

What we recommend, case by case

• Online shop or service: claim the listing with DNS verification, which ties the claim to your domain rather than a phone number or inbox. Complete Persona verification to see insights. Connect a product feed only after reading Section 3 of the terms.
• High-street shop or restaurant outside the US: nothing to claim yet. Keep your details current where ChatGPT can read them (your website, directory listings), since OpenAI says its answers may also draw on other sources.
• Business whose content is the product: if your robots.txt blocks OpenAI's crawlers, decide first whether to keep that block. The listing and insights do not require lifting it; the product feed does.

More practical guides are in the Guides section.

What we don't know yet

OpenAI gives no timetable for opening high-street listings outside the US, nor for Business Agents in Europe. The reporting thresholds are not published, and the help centre only says review time “varies”. The service is in beta, and the policy says OpenAI may update it over time.