Deleted or Not? Court Order and OpenAI’s Silence Leave Users in the Dark How ChatGPT User Data is Handled
- Tomasz Barczyk
- May 30
- 4 min read
A federal court has directed OpenAI to preserve user chat logs that it would otherwise delete, a significant development in the ongoing copyright infringement litigation brought by The New York Times. This order, and OpenAI’s subsequent actions, raise important questions about users’ ability to control the fate of their ChatGPT conversations and about OpenAI’s obligation to clearly communicate how user data is being retained and used.
On May 13, 2025, the U.S. District Court for the Southern District of New York, in In re: OpenAI, Inc., Copyright Infringement Litigation, Case No. 23-cv-11195 (MDL No. 25-md-3143 (SHS) (OTW)), issued an order compelling OpenAI to "preserve and segregate all output log data that would otherwise be deleted on a going forward basis until further order of the Court."[1] This directive means that even if users activate "Temporary Chat" mode or manually delete their ChatGPT conversations, OpenAI must retain these logs for potential discovery in the lawsuit.
OpenAI quickly contested the ruling. The company filed a Motion for Reconsideration on May 14, 2025,[2] arguing that the order would force it to "disregard legal, contractual, regulatory, and ethical commitments to hundreds of millions of people, businesses, educational institutions, and governments around the world." OpenAI asserted that compliance presented "many practical and engineering challenges" and that the data in question was unlikely to be relevant beyond the "tens of billions of user conversations OpenAI has retained." The Court denied this motion on May 16, 2025, though without prejudice.[3] OpenAI subsequently filed a supplemental opposition on May 23, 2025, again urging the Court to reconsider, emphasizing the "unprecedented, unwarranted" nature of the order and the significant technical and financial burdens of wholesale preservation.[4]
Operational Strain and the Silence Around Policy Shifts (if any?)
OpenAI’s court filings reveal its perspective on the significant technical difficulties in implementing the preservation order. The company described the need for "custom engineering solutions" and expressed concerns about real-time implementation when users can instantly delete conversations.[5] During a January 22, 2025 hearing, OpenAI's counsel explained that users can "elect to delete a specific ChatGPT conversation," and that "numerous privacy laws and regulations throughout the country and the world . . . contemplate these type of deletion requests."[6]
Despite the court order fundamentally altering how it must handle some user data, OpenAI has not made public announcements or updated its terms of service, privacy policy, or data retention policy to reflect the new obligation. Under existing policies and user expectations, deleted conversations—whether manually erased or created in 'Temporary Chat' mode—are presumed to be removed from OpenAI’s systems. Yet, users currently receive no indication that such data may now be retained for litigation purposes. This disconnect creates a transparency gap, leaving users unable to make fully informed decisions about their data. This silence creates a transparency gap, leaving users unable to make fully informed decisions about their data.
The legal exchanges in the case, including filings from the News Plaintiffs such as their supplemental memorandum of May 16, 2025,[7] inadvertently shine a light on these operational complexities. The very discussion of preserving data that users might actively try to delete—a scenario the court itself considered with a hypothetical about a user attempting to erase traces of misuse[8]—illustrates the sophisticated systems OpenAI would need to implement and manage to meet the court's preservation demands effectively. These proceedings highlight the tension between broad discovery mandates and the practical realities of managing dynamic user data at scale.
Conflicting Obligations and Transparency Gaps for Users
This situation presents a difficult scenario:
If OpenAI fully complies with the preservation order but continues to indicate to users that their data is "deleted," it misleads users about their data's status.
If OpenAI honors user deletion requests as usual, it violates the court order.
Partial compliance leaves both the court and users uncertain about actual data handling practices, raising the risk that users may unknowingly rely on misleading assumptions about whether their data has truly been deleted. This lack of clarity could expose users to unintended disclosures or legal vulnerabilities, especially if sensitive information is preserved without their knowledge or consent.
The preservation order also impacts businesses using OpenAI's services. For instance, Spark Innovations Corp. filed a motion to intervene on May 19, 2025,[9] objecting to the retention of its "confidential, trade-secret, and competitively sensitive business information" generated through OpenAI platforms. Although the court later denied this specific motion on other grounds,[10] the concerns raised highlight the broader implications for non-party data.
How the Court Order Could Affect Your ChatGPT Data
OpenAI maintains in its filings that its default practice is to retain conversations unless a user expresses a preference otherwise, and that "tens of billions of conversations" are already retained by default.[11] However, the company also argues that the court's order forces it to override user privacy preferences and presents significant technological challenges.[12]
However, the company also argues that the court's order forces it to override user privacy preferences and presents significant technological challenges.
Given OpenAI's arguments about the difficulty of full compliance and the lack of public communication regarding the preservation order, users concerned about the actual deletion of their ChatGPT data face uncertainty. The ongoing litigation will likely bring more clarity. In the meantime, individuals and businesses should assume that any data entered into ChatGPT could be retained, even if deleted from their end. Users should avoid inputting confidential, privileged, or otherwise sensitive information and should consider revisiting any prior conversations they believed were deleted. Reviewing OpenAI's policies periodically for updates and monitoring related court developments are prudent steps until greater transparency is achieved.
[1] ECF No. 551.
[2] ECF No. 558.
[3] ECF No. 559.
[4] ECF No. 578.
[5] ECF No. 558; ECF No. 578.
[6] Transcript of Jan. 22, 2025 Hearing, p. 42, see ECF No. 551
[7] ECF No. 560.
[8] Transcript of Jan. 22, 2025 Hearing, p. 38; ECF No. 560.
[9] ECF No. 566.
[10] ECF No. 586.
[11] ECF No. 558; ECF No. 578.
[12] ECF No. 558; ECF No. 578.