The news, 365 days behind — on purpose Delayed live · replaying 2025

One Year Ago.AI

Remember how fast this is.

04JUN2025replayed
one year on
policyOpenAI · The New York Times

OpenAI slams court order to preserve all ChatGPT logs, including deleted chats

In a copyright lawsuit brought by The New York Times and other news organizations, OpenAI is fighting a court order that it says forces the company to violate user privacy and its own contractual commitments.

OpenAI is fighting a court order that it says forces the company to preserve all ChatGPT user logs, including deleted chats and API data, arguing the order is an unprecedented invasion of privacy that breaks its commitments to hundreds of millions of users.

The order, issued May 13 by U.S. Magistrate Judge Ona Wang in the multidistrict copyright litigation brought by The New York Times and other news organizations, came after the plaintiffs raised concerns that users might delete chats to cover up paywall circumvention. OpenAI says the ruling was premature and based on speculation, not evidence.

In a court filing, OpenAI warns that the order compels it to retain data users explicitly chose to delete, including temporary chats and account deletion requests, violating its privacy policies and potentially exposing sensitive personal and business information. The company says the order imposes a massive engineering burden and puts it at risk of breaching contracts and global privacy regulations.

On Hacker News, where the item has 1,132 points and 932 comments, the thread turns into a debate over the limits of evidence preservation in civil cases. Some commenters argued the order is standard legal procedure, while others warned it sets a dangerous precedent for privacy.

H
Hacker News commenters

Commenters on Hacker News debated the scope of the order, with many comparing it to routine evidence preservation in civil litigation and others expressing alarm at the privacy implications.

One year later — open only if you can handle spoilers

The court later modified the order to allow OpenAI to exclude certain categories of data, but the underlying tension between copyright discovery and user privacy remains unresolved. The case continues to be closely watched as a test of how traditional evidentiary rules apply to AI services.

Replay thisPost on XRedditHNLinkedIn

The Weekly Replay · free by email

This week, one year ago — every Sunday.

One email each Sunday: the week's replayed AI news, with the one-year-later annotations included. Written like it's breaking — dated like it isn't.

Free · double opt-in · unsubscribe anytime · privacy