US Government Sides With OpenAI in NYT Copyright Case

Today the US government picked a side in the defining legal fight of the AI decade. The Justice Department has sided with OpenAI in New York Times v. OpenAI — the flagship lawsuit over whether training LLMs on copyrighted text is fair use. Reuters, TechCrunch, WIRED, The New York Times, The Intercept and AppleInsider all reported it on September 2.

What shipped

Reports describe a filing backing OpenAI on the core question: training on copyrighted material should not be treated as infringement. AppleInsider's read of the administration's position: AI companies can't innovate without this use. The Intercept — itself a plaintiff in the parallel authors' suit against OpenAI — framed the same position as permission to train on its own articles. WIRED's coverage has the telling detail: the administration is siding with "AI giants," plural, meaning the brief's logic reaches well beyond one lab.

Why it matters

Amicus positions from the Justice Department carry unusual weight in district court, and this is the case the whole industry is watching. If the NYT's infringement theory wins, every model trained on the open web is retroactively exposed. If OpenAI's fair-use reading holds, the copyright chapter of the training wars effectively closes in the labs' favor. The timing lands hard too — the same week Mistral quietly made training on consumer data a default and Anthropic paused training over unauthorized actions. The executive branch just staked out its ground; the courts are next.

Verdict: the clearest signal yet that the US is betting the AI economy on fair use. Watch the SDNY ruling — and the author-MDL cases queued behind it.