Breaking

U.S. Justice Department Backs OpenAI Fair-Use Defense in New York Times Copyright Suit

September 2, 2026 at 15:03 EDT

Policy & Regulation
Foundation Models

The U.S. Justice Department has told a Manhattan federal court that training large language models on copyrighted text is generally fair use, intervening for the first time in the wave of lawsuits over how AI systems are built. The filing supports OpenAI in its copyright battle with The New York Times and other newspapers and comes nearly three years after the Times sued.

Sept. 1, 2026 · U.S. Department of Justice

Washington enters the AI copyright fight

The Justice Department told a Manhattan court that training large language models on copyrighted text is generally fair use — the first U.S. government filing in a wave of training-data lawsuits, nearly three years after the case began.

Magnitude

Nearly 3 years before the first federal brief

1 yr

Suit filed, 2023

3 yrs

DOJ files, 2026

The government stepped in only after years of private litigation over ChatGPT training data.

1st

U.S. government brief on AI-training fair use

Statutory fair-use factors the court still weighs

$1.5B

Anthropic piracy settlement after a 2025 split ruling

Billions

Damages sought from OpenAI and Microsoft

How the brief splits the pipeline

Training, not output, is the government's target

1. Collect

Acquire text, including news articles

→

2. Train

Learn patterns — DOJ: extraordinarily transformative

→

3. Output

Chat replies stay a separate legal question

The government's case

Training copies are not shown to the public, so they are not substitutes. Restricting them, officials say, would slow science, raise licensing barriers, and weaken U.S. AI against foreign rivals.

The publishers' case

Newsrooms say unlicensed training on millions of articles lets chatbots compete with journalism. They argue companies can pay, as some licensing deals already show.

What the filing does not do

Advisory only — Judge Sidney H. Stein still decides

The United States is not a party to the suit.
2025 rulings already split training from piracy and market-dilution theories.
No appellate court has settled the AI-training standard.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.