The US government files a brief backing OpenAI on training AI with copyrighted works — in the NYT's own lawsuit

2 sources

The US government filed a 20-page brief in The New York Times v. OpenAI (Case 1:25-cv-03483-SHS-OTW, SDNY) — Document 316, filed September 1, 2026 per the filing's own page stamps — defending OpenAI's unlicensed use of copyrighted works for LLM training. Per the brief as quoted by TechCrunch, it cites President Trump's January 2025 executive order and argues that constraining LLM development 'under a misunderstanding of fair use doctrine would thwart such creative and scientific progress while hindering American prosperity and economic mobility.' The case is pending; no court has ruled. TechCrunch contrasts this with Judge William Alsup's approval of Anthropic's $1.5 billion settlement — approved over shadow-library pirating, not training itself.

Why now

The brief landed Tuesday and TechCrunch's read Wednesday, but the creator-side stakes haven't been walked yet: this intervention lands just days after music publishers sued Anthropic — with the government now arguing training can be fair use while that case and the NYT's own proceed. For audiences who just watched Sony and Warner Chappell sue Anthropic, the one-two of 'a publisher sues' followed by 'the US government argues training is fair use' is the clearest window all year to explain what fair use actually decides — and the brief's own filing date (which TechCrunch's report omits; AITopic read it from the PDF stamps) is a detail most coverage will miss.

Why it is worth publishing

Highest-stakes creator-economy story of the day with an unusually strong evidence chain: AITopic verified the brief's existence, page count, case number, and filing date directly from the court PDF's own stamps, with TechCrunch's full read carrying its content. The 'filing is not a ruling' line is the trust angle most coverage will fumble.

Evidence basis

Government brief PDF fetched on 2026-09-03 (document-level facts read from page stamps: 20 pages, Document 316, filed 09/01/26, Case 1:25-cv-03483-SHS-OTW) + TechCrunch full read on 2026-09-03 (posted 10:09 AM PDT September 2, 2026).

“In a court filing, the US government says training AI on copyrighted works can be fair use.”

Angle

Teach the procedural stack, not just the headline: what a government brief is (an interested party's litigation position, not a ruling), where it sits in NYT v. OpenAI, what fair use actually decides for training — and what it does not decide (the publishers' suits, including the music publishers v. Anthropic filing, continue regardless).

Format

Carousel

Demo idea

Five-slide ladder: slide 1 what was filed (20 pages, September 1, by the US government — show the PDF stamp); slide 2 the fair-use quote; slide 3 what a brief can and cannot do; slide 4 the Alsup line from the Anthropic settlement (piracy vs training); slide 5 what creators should watch next in the docket.

Platform notes

Say 'filed a brief arguing' every time — the case is pending and no court has ruled; 'sides with OpenAI' is TechCrunch's characterization. The two brief quotes are carried by TechCrunch's read of the filing, and the filing date comes from the PDF's own stamps (AITopic read them directly). The Anthropic settlement comparison is TechCrunch's framing of a different case.

Usable claims

The US government filed a 20-page brief — Document 316, filed September 1, 2026 per the filing's own page stamps — in The New York Times v. OpenAI (Case 1:25-cv-03483-SHS-OTW, US District Court for the Southern District of New York), defending OpenAI's unlicensed use of copyrighted works for LLM training. Per the brief as quoted by TechCrunch, it cites President Trump's January 2025 executive order 'Removing Barriers to American Leadership in Artificial Intelligence' and argues: 'The United States has a strong interest in continuing to develop a robust and competitive artificial intelligence industry that sets the standard for the practice and procedure of AI use globally', and that 'Constraining LLM development under a misunderstanding of fair use doctrine would thwart such creative and scientific progress while hindering American prosperity and economic mobility.'
Context, per TechCrunch: The New York Times and other publishers argue it is illegal for AI companies to train models on their copyrighted material — the brief is the government's intervention on OpenAI's side of that dispute. TechCrunch also draws the line it sees in prior case law, citing Judge William Alsup's approval of Anthropic's $1.5 billion copyright settlement with writers — approved over pirating books via shadow libraries, not over training itself — and quoting Alsup that LLMs trained on copyrighted works do so 'not to race ahead and replicate or supplant them — but to turn a hard corner and create something different.' Per TechCrunch's own accounting of gaps: the article does not identify which agency authored the brief, does not give the filing date (AITopic read it from the PDF stamps as September 1, 2026), and records no reaction from the Times, OpenAI, or creators' groups.

Evidence pipeline

From the news

Breakdown

The US government filed a 20-page brief (Document 316, filed September 1, 2026 per the PDF's own stamps — a detail TechCrunch's report omits and AITopic read directly) in NYT v. OpenAI, defending unlicensed training on copyrighted works and citing the January 2025 executive order. This breakdown holds the line between arguing and deciding: the case is pending, 'sides with OpenAI' is TechCrunch's characterization, the brief's quotations are carried by TechCrunch's read of it, and the Alsup/Anthropic-settlement contrast is a different case about pirating, not training.

Sources

Risks

Say 'filed a brief arguing' every time, name the case and court, and state on the record that the case is pending and no court has ruled — the brief is one party's advocate (the US government) taking a position.
Attribute quoted lines as 'the brief says, as quoted by TechCrunch' where practical, never add 'quotes' beyond the two the source carries, and keep the filing-date fact tied to the PDF stamps.

Demo ideas

Stamp-read clip: open the CourtListener PDF on screen, zoom on 'Document 316 Filed 09/01/26', and explain what a docket stamp proves
'Brief vs ruling' explainer: who files briefs, what they argue, what judges do with them — using this filing as the live example