Business feeds need a data-rights receipt
In short: OpenAI's Business Hub feed terms change data-use permissions, so merchants need a scoped approval and a verified exit receipt.
On 10 November 2026, OpenAI launched Businesses in ChatGPT for a service it says reaches 1.2 billion weekly users. Eligible merchants can claim a profile, upload product feeds, see aggregated mentions and website clicks, and publish a Business Agent.
The integration is also a data-rights decision. Business feeds need a data-rights receipt because OpenAI's current terms couple catalogue submission with crawler access and a specific text-and-data-mining commitment. Treating the upload as routine marketing plumbing leaves the consequential permission outside the approval record.
OpenAI's Business Hub couples product data with access terms
The launch announcement says retailers can submit pricing and availability, while eligible US ecommerce and online-service businesses can build an agent that answers product and policy questions. OpenAI also says Business Agent chats are labelled and separate from ChatGPT's independent answers. Businesses receive aggregated insights, not people's conversations.
The Businesses in ChatGPT terms add the part an integration ticket can miss. A Product Feed provider must permit OpenAI and its affiliates' crawlers and agents to access and interact with relevant sites. For the Customer Data supplied to the service, the provider also agrees that text-and-data-mining opt-outs or similar reservations have no effect.
That wording has boundaries. Customer Data includes the feeds, business information, media, links, linked material, guidelines and other material made available to the service. It does not support the broader claim that connecting one feed relicenses every page the merchant owns. The same terms let OpenAI decline or remove feed content and continue to maintain a business listing from information it is otherwise entitled to use.
Business feeds need a data-rights receipt before upload
I have not connected a merchant feed or tested crawler and deletion behaviour for this article. This is labelled editorial judgment, not a Trigent or client deployment.
Before a team switches on Business Hub, I would put one receipt beside the feed. It would name the approved domains and data classes, exact feed version, business, legal and technical owners, crawler and agent access accepted, intended downstream uses, withdrawal path and evidence that a requested removal propagated.
The receipt is deliberately dull. Its job is to stop a product catalogue, a crawler permission and a data-use commitment from passing through three different teams with no shared owner. It also gives procurement something concrete to compare if the merchant later distributes the same catalogue through several agent platforms.
This extends the archive's data-rights clock. A citation licence can expire while a model still remembers an old fact. Here the first question comes earlier: which business data did we deliberately make available, under which terms, and how will we prove the relationship ended?
OpenAI's terms do not promise placement or sales
OpenAI's business policy says it does not verify every submitted detail or resolve every factual dispute. Its product page separately says claiming a business does not guarantee higher placement or sales.
So the receipt is not a performance dashboard. It cannot prove that the agent improved discovery, conversion or data quality. Keep commercial measurement in a different record: mentions, clicks, downstream sessions, accepted orders and corrections. Mixing those outcomes into the permission receipt makes both harder to audit.
What's in it for you
- Route the next product-feed connection through an explicit data-rights approval.
- Keep crawler scope and commercial performance in separate evidence records.
- Test withdrawal and removal before the feed becomes operationally hard to unwind.
A business feed is publishable data under terms, not merely rows moving between two catalogues.
