OpenAI Rewrites Businesses in ChatGPT Terms After Lawyers Flag Broad Claims Release

OpenAI Rewrites Businesses in ChatGPT Terms After Lawyers Flag Broad Claims Release

geoNovember 20, 2026
By Antonio Fernandez

TL;DR

  • OpenAI updated its Businesses in ChatGPT Terms on 18 November 2026, eight days after first publishing them on 10 November.
  • Digiday reported on 20 November that lawyers read the original Section 10 as asking businesses to give up claims, possibly including claims about content on their sites before sign-up.
  • The 18 November text ties the release to claims arising from a business's Business Agent (Section 5.2); a Davis+Gilbert partner said it no longer reaches unrelated disputes.
  • Section 4.1 lets OpenAI display a listing whether or not a business claims it, and Section 4.2 makes the licence over name, address, hours and website perpetual and irrevocable.

OpenAI changed its Businesses in ChatGPT Terms on 18 November 2026, eight days after publishing them, and the current version narrows the legal release a business gives OpenAI, tying it to a Business Agent the business enables or publishes. Under the 18 November text, the release in Section 5.2 covers claims "arising from your Business Agent, including any use of Customer Data with the Business Agent." Digiday reported on 20 November that lawyers had read the earlier wording as asking businesses to give up a wider set of claims, possibly including claims about content that sat on their websites before they ever signed up. Several other clauses that matter to brands and their agencies remain in the current text, and they are worth reading before anyone clicks "claim".

This article is a plain-language summary for marketers, written from OpenAI's published terms and from press reporting. It is not legal advice. Any business that plans to claim a ChatGPT listing or publish an agent should have its own lawyer read the full terms and the underlying agreement they sit on.

What changed in the Businesses in ChatGPT Terms on 18 November

OpenAI's Businesses in ChatGPT Terms now carry the line "Last updated: November 18, 2026". According to Digiday's Sara Guaglione, who reported the story on 20 November, the terms were first published on 10 November, the day OpenAI launched Business Hub and Business Agents in beta. Digiday said the rewrite came hours after it first contacted OpenAI on 18 November.

Digiday described the original problem this way. Lawyers said Section 10 of the first version appeared to require businesses to give up certain legal claims against OpenAI, and that the wording could reach claims about content published on a business's website before it signed up. The original also carried a mandatory arbitration provision with a 30-day opt-out window, a class-action waiver, and a separate "covered claims" provision. Alan Chapell, a privacy expert and founder of Chapell and Associates quoted by Digiday, said publishers who accepted the original terms could have faced restrictions on how they pursued such claims against OpenAI. Digiday added that this could have limited publishers' ability to bring copyright infringement claims against OpenAI down the road.

An OpenAI spokesperson told Digiday that the updated terms replace the separate "covered claims" provision with language in Section 5.2, "Customer Responsibilities", that ties the release to claims arising from a business's Business Agent. An OpenAI spokesperson told Digiday the earlier version was "not intended to cover this scenario" and said the company keeps reviewing its terms to make them as clear as possible. Gary Kibel, a partner at the law firm Davis+Gilbert, told Digiday that the revised text still requires businesses to waive certain claims connected to Business Hub and Business Agents, but that it no longer extends to unrelated disputes with OpenAI. Kibel also described the original language as deliberate and aggressive.

What the current page does not contain

The 18 November page has no arbitration clause and no class-action waiver of its own. Digiday did not say whether those provisions were removed, moved or left in another document. The terms say they "supplement and are incorporated into the Agreement", meaning whatever master terms a business already has with OpenAI, and that agreement is not reproduced on the page. So the absence of arbitration language on this page does not by itself tell a business how disputes would be handled. That is a question for the full contract set, not for this summary.

The clauses in the 18 November text, side by side

The table below lists the provisions most relevant to marketers, quoted or paraphrased from the 18 November version of the terms. Section numbers are OpenAI's.

The clauses in the 18 November text, side by side
SectionWhat the 18 November text says
1.2 Tool-Specific TermsOpenAI may develop, configure and test Tools, including Business Agents, for a business before it elects to use them.
3 Product FeedA business that supplies a product feed must let OpenAI's crawlers and agents access its websites, and any text and data mining opt-out it reserves "will have no effect" against OpenAI.
4.1 Business ListingsOpenAI may create and display a listing "regardless of whether you claim, manage, or cease managing it", and claiming does not give exclusive control over its contents.
4.2 Public InformationThe display licence for basic business information (name, address, contact details, category, opening hours, website) is perpetual and irrevocable.
5.2 Customer ResponsibilitiesA business that enables or publishes an agent is responsible for its content, Actions and Outputs "whether or not you provided, reviewed, or approved them", and irrevocably releases OpenAI from claims arising from the agent.
5.3 Distribution and PromotionPublishing grants users an irrevocable, royalty-free licence to Customer Data that appears in the agent's Outputs.
10 Continued Provision of ServicesTermination does not require OpenAI to modify, delete or stop offering products in which Customer Data has been used, and the business waives equitable remedies including injunctive relief.

The listing exists whether or not a business claims it

Section 4.1 is the clause that changes the usual logic of "should we claim this profile?". OpenAI says it may create, maintain and display a business listing on its properties using Customer Data and "other information OpenAI is otherwise entitled to use", regardless of whether the business claims it, manages it or stops managing it. Claiming gives the business a way to submit and correct details, subject to identity and authority checks. It does not give exclusive control: OpenAI says it may supplement, edit or replace information from other sources, or decline to display what the business submits. The section also says OpenAI does not guarantee verification.

Two more details follow. By using Business Hub, a business agrees that OpenAI's AI assistant may phone it on behalf of ChatGPT users, with an opt-out link on the page. And Section 4.2 makes the display licence for basic public information perpetual and irrevocable. In practice (this is RA's reading, not OpenAI's wording), a business that claims a listing and later walks away cannot expect its name, address, hours and website to be withdrawn from display because it stopped participating.

Business Agents: built for you, released by you

Section 5.1 says OpenAI "may create and configure a Business Agent for your business before you elect to use it." Creating, configuring, previewing or linking an agent does not publish it; an agent reaches users only if the business chooses to publish. The same section says publishing does not authorise paid advertising, which needs a separate advertising relationship under OpenAI's Advertising Terms.

Once a business enables or publishes an agent, Section 5.2 puts responsibility for the agent on the business, including content, configuration, safety testing, Actions and Outputs, "whether or not you provided, reviewed, or approved them." The business must correct issues within its control and tell OpenAI about inaccuracies or unauthorised behaviour. The release then follows: the business, for itself and its affiliates, irrevocably releases OpenAI and its affiliates from claims arising from the agent, including any use of Customer Data with it.

Section 5.6 sets the other side of that bargain. OpenAI says agents may produce inaccurate or unexpected responses, does not warrant that an agent will follow instructions, and may modify, suspend or remove an agent at any time without notice. Section 9 adds an indemnity: the business defends OpenAI against third-party claims tied to its Customer Data, its products, its listing or agent, and allegations such as false or misleading advertising or consumer protection violations.

Why this combination matters (analysis)

Read together, the clauses put the business on the hook for outputs from an agent that OpenAI may have built first, and that OpenAI says may not follow instructions. The 18 November narrowing limits the release to the agent itself, which is a real change from a release that lawyers thought could reach unrelated disputes. It does not change who answers for what the agent says. For a brand, the practical consequence is that publishing an agent is a decision for legal and compliance teams as well as for the marketing team running the listing.

Product feeds and crawler access

Section 3 applies to any business that supplies a product feed. It requires the business and anyone acting for it to let OpenAI's and its affiliates' crawlers and agents access and interact with its websites under OpenAI's documentation. It also says the business and its affiliates do not exercise or reserve any text and data mining exception or opt-out against OpenAI, and that if they do, those rights "will have no effect" with respect to OpenAI. The section adds that OpenAI may decline or remove feed content at its discretion, and that personal data in a feed must have been publicly available in mass media.

For SEO and GEO teams, this clause is the one that connects the terms to everyday technical decisions. A company that blocks some AI crawlers in robots.txt, or that has published a text and data mining reservation, should check whether supplying a feed to Business Hub conflicts with that stance. That is an internal policy question the terms do not answer for you.

What is new compared with RA's 11 November coverage

Relevant Audience covered the launch itself on 11 November, in "OpenAI Launches Businesses in ChatGPT: Business Hub Claims, Mention Data and Business Agents". That piece dealt with how claims and verification work, how Business Hub counts mentions and clicks, and how Business Agents connect to ChatGPT Ads. It did not cover the legal terms. Three things are new since then:

  • The terms themselves became a story. Lawyers quoted by Digiday on 20 November read the 10 November version as asking businesses to give up a broad set of claims.
  • OpenAI rewrote them on 18 November, removing the separate covered-claims provision and tying the release to the Business Agent in Section 5.2.
  • The current text makes clear that a listing can exist without a claim, that basic business information is licensed perpetually, and that OpenAI may build an agent before the business opts in.

Earlier still, on 7 November, RA covered reports that OpenAI was building a "Claim your business in ChatGPT" hub, before any launch or terms existed.

What the sources did not say

Neither OpenAI's terms page nor Digiday's report gives a full redline of the 10 November and 18 November versions. Digiday described the changes and showed screenshots, but the earlier text is not published on OpenAI's page. Neither source says whether the arbitration and class-action provisions still apply through another document. Neither says how many businesses accepted the 10 November version, or whether acceptance of the earlier text carries over in any way. Neither mentions Thailand, and the terms page does not list which countries Business Hub is available in. The spokesperson quote addresses intent; it does not concede that the earlier text had the effect lawyers described.

What this means for Thai marketers

None of the sources refer to Thailand, so what follows is RA's analysis for brands and agencies working in the Thai market. Claiming a ChatGPT business listing can look like a routine local-visibility task, similar to a Google Business Profile. The terms show that it carries more legal weight than a profile edit. Before anyone claims a listing for a client, an agency should have written sign-off from someone authorised to bind the business, because Section 7 makes the person claiming represent that they have that authority.

A short checklist, offered as a starting point for a conversation with counsel rather than as advice:

  1. Confirm who in the business may accept the terms and the underlying Agreement on its behalf.
  2. Decide separately whether to publish a Business Agent. Claiming a listing and publishing an agent carry different obligations under Sections 4 and 5.
  3. Review default settings after claiming. Section 2.3 says some functionality may be enabled by default and that the business is responsible for settings it leaves enabled.
  4. Check the product feed clause against the company's crawler and text and data mining policy.
  5. Check how Lead Information from an agent would be handled. Section 5.5 puts compliance with privacy law on the business, which for a Thai company means reviewing it alongside its duties under the Personal Data Protection Act.

Teams already planning ChatGPT SEO work, or broader generative engine optimisation, should treat the listing decision as part of that strategy. Businesses with physical locations will recognise the pattern from local SEO, where profile ownership and data accuracy carry most of the weight.

Frequently asked questions

What did OpenAI change in the Businesses in ChatGPT Terms?

OpenAI narrowed the release businesses give it, so that in the 18 November 2026 text it covers claims arising from a business's Business Agent, including use of Customer Data with that agent. An OpenAI spokesperson told Digiday that a separate "covered claims" provision from the 10 November version was replaced, and a lawyer it quoted said the revision no longer reaches unrelated disputes with OpenAI.

Does OpenAI create a ChatGPT listing for a business that never claims it?

Yes, Section 4.1 says OpenAI may create, maintain and display a listing whether or not the business claims, manages or stops managing it. Claiming lets the business submit and correct details, but OpenAI says it may still edit or replace information from other sources.

Can a business withdraw its information after claiming a listing?

Not fully for basic details, because Section 4.2 makes the display licence for name, address, contact details, category, opening hours and website perpetual and irrevocable. Section 10 also says termination does not require OpenAI to delete products in which Customer Data has been used.

Is Business Hub available in Thailand?

The terms do not say. The page lists no countries, and neither OpenAI's terms nor Digiday's report mentions Thailand, so a Thai business should check eligibility inside Business Hub itself.

Is this article legal advice?

No, it is a marketing summary of public documents and press reporting. A business should ask its own lawyer to review the full terms and the underlying OpenAI agreement before claiming a listing or publishing an agent.

Bottom line

The 18 November rewrite removed the broadest reading of the original release, and that is a real improvement for any business weighing Business Hub. The rest of the bargain still stands: a listing that exists without your consent, a perpetual licence over basic details, an agent OpenAI may build before you opt in, and responsibility for that agent's outputs once you publish it. Brands that want ChatGPT visibility can still pursue it; they should go in with legal review done first. Relevant Audience helps businesses plan ChatGPT and AI search visibility and can work alongside a company's legal advisers on what to claim and when.

Antonio Fernandez

Antonio Fernandez

Founder and CEO of Relevant Audience. With over 15 years of experience in digital marketing strategy, he leads teams across southeast Asia in delivering exceptional results for clients through performance-focused digital solutions.

Share to:
Copy link:

Read us often? Add Relevant Audience as a preferred source so our articles surface more in your Google results.