The Retention Order Ended; a Limited Historical Hold Remains
OpenAI’s October 22, 2025 update says its earlier obligation to retain consumer ChatGPT and API content indefinitely ended on September 26, 2025. The company returned to its standard retention practices. A limited set of historical user data from April–September 2025 remains under legal hold. This June 2025 report has been corrected to distinguish that historical order from OpenAI’s later update; it does not establish any subsequent court ruling. Related coverage: AI.
What the June 2025 Notice Said
In its June 5, 2025 notice, OpenAI said the New York Times and other plaintiffs sought preservation of consumer ChatGPT and API content that would otherwise be removed. The notice described a preservation order and OpenAI’s challenges to it. It did not say every conversation from every product was covered. This report relies on the company’s published account rather than an independent reading of the court docket. Related coverage: OpenAI.
Deletion, Retention and Training Are Different Controls
The original article incorrectly treated ChatGPT history, model training and deletion as one setting. OpenAI’s published policy description says a deleted chat is removed from the user’s account immediately and scheduled for deletion from its systems within 30 days, subject to legal or security exceptions. Turning off training does not itself create a zero-retention service. Enterprise and Edu were excluded from the preservation order described in the June notice; that exclusion was not a blanket promise that those products retain no data.
OpenAI’s Challenge and the Later Update
OpenAI said it sought reconsideration and appealed the preservation order. Its October 22 update subsequently reported that obligations under the earlier order ended on September 26. The older FAQ remains on the same source page, but its statements about indefinite retention going forward describe the earlier situation. The dated update governs how this article describes that order. Unsupported quotations and predictions about the appeal have been removed.
What Remained Under Legal Hold
The October update says OpenAI would securely retain limited historical April–September 2025 user data while the New York Times continued to seek its preservation. It says the company was no longer required to retain new data indefinitely going forward, or conversations originating from the European Economic Area, Switzerland or the United Kingdom. This is OpenAI’s description of that update, not an independent determination of every user’s legal position or the eventual outcome of the lawsuit.
Which Products Were Covered by the Earlier Order?
The June notice identified ChatGPT Free, Plus, Pro and Team, and API use without a Zero Data Retention agreement, as affected. It excluded ChatGPT Enterprise and Edu and API customers using ZDR endpoints under the relevant amendment. Those product-specific exceptions matter: an API account is not automatically a ZDR account, and an Enterprise or Edu exemption from this order does not mean every workspace has zero retention. Administrators, endpoint choices and configurations affect ordinary retention.
Who Could Access the Preserved Data?
OpenAI says the data under legal hold is stored separately and may be accessed only by a small, audited legal and security team to meet legal obligations. Its October update says the retained historical data would not be turned over to the New York Times, the court or anyone else at that time. Preservation and disclosure are separate issues; the existence of a hold does not mean plaintiffs received the data or that it was used for model training.
What the Published Notice Says About API Data
The October update summarizes ordinary API deletion as occurring after 30 days. The same page’s policy details distinguish logs from application state: business customers control application-state retention through endpoints and configuration, while API inputs and outputs are removed from OpenAI logs after 30 days unless legally required otherwise. It describes ZDR endpoints as not logging inputs and outputs or retaining them for application state. These distinctions should be checked against the current documentation for the endpoint being used; this article does not certify a particular integration’s retention settings.
What This Report Does and Does Not Establish
The source establishes OpenAI’s dated public account of the earlier order and the October 2025 change. It does not establish later litigation outcomes, automatic consequences for other AI providers, new legislation or a general legal privilege for chatbot conversations. The original article’s anonymous expert quotation and predictions of an inevitable industry-wide hold were unsupported and have been withdrawn.
Practical Reading of the Update
The earlier advice to assume no ChatGPT content could ever be deleted was too broad. Read the product’s current deletion and retention controls, distinguish a historical legal hold from ordinary storage, and verify any ZDR arrangement rather than inferring it from a subscription name. Enterprise is not an assurance of “airtight confidentiality.” This article does not offer legal advice or a hands-on audit of an account, workspace or API deployment. Related coverage: ChatGPT.
Sources
Primary source: OpenAI’s June 5, 2025 notice and October 22, 2025 update. Source page observed October 8, 2026; the dated notice is historical evidence, not an independent verification of subsequent developments. The original secondary links below are retained as historical reporting.
- The Verge – “OpenAI is storing deleted ChatGPT conversations as part of its NYT lawsuit” (theverge.com)
- Gizmodo – “OpenAI Appeals ‘Sweeping, Unprecedented Order’ Requiring It Maintain All ChatGPT Logs” (gizmodo.com)
- The Decoder – “OpenAI starts retaining all ChatGPT user data, including deleted chats and API data” (the-decoder.com)
The Kingy Brief
Get The Kingy Brief.
AI changes, original tests and one practical thing to try. Fridays at 09:00 Vancouver time.
Free · Double opt-in · Unsubscribe anytime
Signup help and newsletter schedule
Signup form provided by Beehiiv. After submitting, check your inbox for "Confirm your subscription to The Kingy Brief" and open its confirmation link. Check Spam or Promotions if you cannot find it.
Fridays at 09:00 Vancouver time: source-checked AI changes, original tests and one practical thing to try. The weekly restart begins October 9, 2026. We skip a week when there is not enough verified material. Free. Unsubscribe anytime.