Businesses in ChatGPT Terms: Why Our Law Firm Paused Before Getting Listed
The short answer: We want Antonelli Law® listed in ChatGPT so people who need a copyright defense lawyer can find us, which is why we read the Businesses in ChatGPT Terms. Those terms (last updated November 9, 2026) ask a business to do much more than license its name and logo.
Section 10 asks you, and every company affiliated with you, to promise never to sue OpenAI over your submitted data or over content published on your own websites, to release those claims “known or unknown,” and to give up injunctive relief.
For a law firm that must protect prospective client confidentiality and its own copyrighted work, that is a lot to sign for a directory listing. So we asked OpenAI Legal to strike or narrow Section 10 for our listing.

This week makes questions about the Businesses in ChatGPT Terms more pressing. Google’s threat intelligence team reports that attackers now use AI to turn newly published patches into working exploits faster than before, and organizations are racing to patch. When a vendor’s terms say it may keep and use your data indefinitely and you cannot ask a court to stop it, that vendor’s security becomes part of your clients’ risk.
What are the Businesses in ChatGPT Terms?
The Businesses in ChatGPT Terms govern OpenAI’s Business Hub, where a business can create, claim, and manage a listing in ChatGPT, plus optional tools such as a Product Feed and an AI “Business Agent.” You accept them by “enabling, accessing, or using Businesses in ChatGPT, or by providing data or materials through it.” They sit on top of a separate underlying “Agreement” with OpenAI.
Some of the Businesses in ChatGPT Terms are ordinary for an online directory. A display license for your name, address, phone, hours, and website is expected. OpenAI’s statement that it guarantees no placement or traffic (Section 8) is standard. And a Business Agent is not published unless you choose to publish it (Section 5.1). Other provisions of the Businesses in ChatGPT Terms go much further.
Which clauses in the Businesses in ChatGPT Terms should a company read twice?
| Section | What it says (plain language) | Why it matters to a law firm |
|---|---|---|
| 2.2 and 11, “Access Purpose” | You provide Customer Data for the “Access Purpose,” defined as “OpenAI’s and its Affiliates’ development, provision, and improvement of their products or services.” | Your data is not limited to powering your listing. |
| 4.1, Business Listings | OpenAI may create and display a listing “regardless of whether you claim, manage, or cease managing it,” and may “supplement, edit, or replace” what you submit. Using Business Hub also means “OpenAI’s AI assistant may make calls to you on behalf of our users” (an opt-out link is provided). | You do not fully control what the listing says. |
| 4.2, Public Information | The display license for basic business information is “perpetual and irrevocable.” | Leaving the platform does not end that license. |
| 5.2 and 5.6, Business Agent | If you publish an agent, you are responsible for its outputs “whether or not you provided, reviewed, or approved them,” and OpenAI does not warrant they will be accurate or lawful. | An AI speaking to prospective clients in a firm’s name raises supervision and advertising-rule concerns. |
| 7(d) and 9, Warranties and Indemnity | You warrant that the facts you supply or approve are accurate and substantiated, and you indemnify OpenAI against claims about your listing or agent, including false-advertising claims. | The firm carries the risk for content it does not fully control. |
| 3, Product Feed | If you provide a feed, any text and data mining opt-out you or your affiliates reserve “will have no effect” as to OpenAI. | A robots.txt or similar opt-out would not protect your sites from OpenAI. |
| 10.1, Covenant Not to Sue | You and your Affiliates “irrevocably and perpetually” agree not to bring any “Covered Claim” against OpenAI, its Affiliates, or their customers and end users. | This reaches far beyond the listing. |
| 10.2, Release | You release all Covered Claims “whether known or unknown,” and waive California Civil Code Section 1542 and similar laws. | Claims you have not discovered yet are included. |
| 10.3, Continued Provision of Services | OpenAI never has to “modify, delete, or cease providing” products that used your Customer Data, and you waive injunctive relief. | Even after you leave, your data may stay in use. |
| 11, “Covered Claim” | Claims relating to Customer Data, to content “published or made available at any time” on websites you or your Affiliates own, or to other content you have rights in that appears online. Claims that OpenAI breached these Terms are excluded. | Your whole web library, past and present, is in scope. |
To be fair to OpenAI, the “Covered Claim” definition in the Businesses in ChatGPT Terms does carve out claims that OpenAI breached these Terms, and the Prohibited Data rule exists to keep sensitive information off the platform. The concern is the reach of Section 10, not the listing itself.
What does Section 10 of the Businesses in ChatGPT Terms mean for a firm’s content and IP?
A company’s websites are its library: years of guides, FAQs, articles, and videos. Read literally, Section 10 would seem to cover claims relating to “any other content, data, or materials that are or were published or made available at any time” on sites owned by the firm or its Affiliates. “Affiliate” means any entity that controls, is controlled by, or is under common control with the firm, so related companies can be bound too.
We help people facing copyright claims every day, and we also care about our own copyrights. Giving up claims over our entire published body of work under the Businesses in ChatGPT Terms, including unknown claims, is not a fair trade for a free listing.
How does this week’s cyber news connect?
We want to be careful here, because accuracy matters.
- On November 16, 2026, Google Threat Intelligence Group (GTIG) and Mandiant published their AI Threat Tracker. It reports that threat actors are using LLMs and agentic frameworks “to fully automate the reverse engineering of patches via differential analysis,” that this can “dramatically reduce the window between patch disclosure and weaponized exploitation,” and that in the past quarter GTIG identified threat actors using LLM frameworks to fully automate vulnerability discovery and exploitation “without a human in the loop for the first time.” GTIG also notes that the recent PHP vulnerabilities were discovered using LLM-enabled analysis.
- CISA urged organizations on October 21, 2026 to immediately patch two critical PHP flaws (CVE-2026-109618 and CVE-2026-109619) that allow unauthenticated remote code execution.
- On November 17, 2026, CNN reported that NSANet, the NSA’s classified internal network, was down, according to six people familiar with the matter. The NSA said it “experienced temporary degradation during scheduled routine maintenance.” Director of National Intelligence Jay Clayton said the outage resulted from patching major software vulnerabilities discovered last month, which CNN identified as the PHP flaws. CNN reported that sources saw no evidence of a cyberattack, and that there was no indication the NSA, FAA, or Microsoft outages were caused by one.
To be sure, on November 18, 2026, Director of National Intelligence Jay Clayton reported that the recent string of tech failures is not a foreign-state cyberattack (White House release; video):
“This is not a cyberattack, and this is not a coordinated effort by a nation state. We’ve looked very closely. Anytime we see something like this, those are the first questions that we ask on behalf of the American people, and we go across and we say, ‘Is that what’s going on here?’ That is not what’s going on here.” Director of National Intelligence Jay Clayton
So the NSA outage was not, on what has been reported, an attack. The lesson is different and still important: patching is hard even for the best-resourced organizations, and Google reports that attackers are using AI to exploit the gap between a patch’s release and its installation. Every company that holds data, AI companies included, is under that pressure.
That is why the Businesses in ChatGPT Terms matter now. If a vendor can keep and use your data indefinitely (Section 10.3), and a claim “relating to” that data, other than one for breach of the Terms, is one you have promised not to bring (Sections 10.1 and 11), then your options after a problem are limited. OpenAI’s business-customer Services Agreement, one of the underlying agreements these terms can sit on top of, also provides for mandatory arbitration, bars class actions, and caps each party’s liability at what the customer paid in the prior twelve months, which for a free listing may be very little.
Frequently asked questions
Do I have to accept the Businesses in ChatGPT Terms to appear in ChatGPT?
No, not necessarily. Under the Businesses in ChatGPT Terms, Section 4.1 says OpenAI may create and display a business listing “regardless of whether you claim, manage, or cease managing it.” Accepting the terms mainly gives you a way to claim and manage the listing.
What is a “Covered Claim” in the Businesses in ChatGPT Terms?
It appears to be a claim, other than a claim that OpenAI breached the Terms, relating to your Customer Data, to content published at any time on websites you or your Affiliates own, or to other content you have rights in that appears online.
Sections 10.1 and 10.2 are made “on behalf of yourself and your Affiliates,” and “Affiliate” includes entities under common control. Related businesses may be covered.
Can I stop OpenAI from using my data after I leave?
Section 10.3 of the Businesses in ChatGPT Terms says termination does not require OpenAI to modify, delete, or stop providing products in which Customer Data was used, and you waive injunctive relief on that point.
Should a business publish a ChatGPT Business Agent?
Each company must decide for itself whether to publish a Business Agent under the Businesses in ChatGPT Terms. Under Section 5.2 the business is responsible for the agent’s outputs whether or not it reviewed them, and Section 5.6 says the agent may produce inaccurate responses. Your own consultation with corporate legal counsel is warranted.
Is it safe to describe my legal case to an AI chatbot?
The Businesses in ChatGPT Terms do not change this: we recommend sharing case details only with a lawyer through a confidential channel.
This page is general information. It is not legal advice, and reading it does not create an attorney-client relationship. Every case is different; results depend on the specific factual and legal circumstances of that case. Quotations from OpenAI’s terms are from the version last updated November 9, 2026. Antonelli Law® defends people and businesses against copyright claims and does not represent copyright holders. Attorney advertising. Information is current as of November 18, 2026.

