Alden Newspapers Amend OpenAI Copyright Suit Versus Microsoft

Eight daily newspapers owned by Alden Global Capital filed an amended US complaint on 21 August 2026 (16:49 GMT) accusing OpenAI and Microsoft of infringing copyrights by using articles to train ChatGPT and Copilot without permission or payment. The publishers’ line in the filing is that the case “arises from defendants purloining millions of the publishers’ copyrighted articles.”

What the amended OpenAI copyright complaint says

MLex reported the amendment at 16:49 GMT on 21 August 2026 (23:49 WIB). This is not a new caption from scratch. It is an amended complaint in an existing US action, now restated against both OpenAI and Microsoft.

The accusation is specific enough to quote and short enough to keep clean. The newspapers say the companies used their articles to train ChatGPT and Copilot without permission or payment. The quoted sentence begins: “This lawsuit arises from defendants purloining millions of the publishers’ copyrighted articles...” That is the publishers’ language, not a court finding. “Purloining” is their word. “Millions” is their count. Do not upgrade either into a verdict.

Microsoft is in the caption because Copilot is in the training claim. If you already cover Copilot’s product surface, keep the two files separate. A URL change is a product story. An amended copyright complaint is a rights story. The shared noun is Microsoft. The facts are not shared.

Why prompt teams should not launder a pleading into a skill

Training-data suits are now a standing weather system around OpenAI. This amendment adds eight Alden dailies and restates the ChatGPT-plus-Copilot theory. It does not tell you to stop calling the API. It does tell you not to paste a paywalled article into a prompt and call that “research.”

A skill that summarizes news should name the outlet, the date, and whether the text was licensed to you. If the user did not provide the article, do not fetch a copyrighted body and stuff it into context as if that were fair by default. Fair use is a defense, not a feature flag.

OpenAI is also in a different fight with itself on safety process, from a training slowdown to a preparedness-team shakeup. Those are not this case. Do not merge copyright, cyber evals, and org charts into one Slack thread titled “OpenAI trouble.”

How to report the amendment without inventing the rest

MLex’s public note is thin. It names eight Alden dailies, OpenAI, Microsoft, ChatGPT, Copilot, an amended US complaint, and the purloining sentence. It does not name the eight titles in the facts used here. It does not name a damages figure. It does not name a court or a case number in the brief we are writing from. Leave those blank. “Eight daily newspapers owned by Alden Global Capital” is the accurate subject.

If you need a client-safe sentence, use this: On 21 August 2026, eight Alden-owned daily newspapers filed an amended US complaint accusing OpenAI and Microsoft of using their articles to train ChatGPT and Copilot without permission or payment. Anything beyond that is extra.

If a skill cites a newspaper, cite the outlet and the date and stop. The citation sentence is Alden newspapers, 21 August 2026, 16:49 GMT, amended US complaint, OpenAI copyright, Microsoft, ChatGPT, Copilot, no permission, no payment, “purloining millions.”

Sources

• Newspapers file amended US copyright claims against OpenAI, Microsoft — MLex, 21 August 2026