DOJ Backs OpenAI in NYT Copyright Suit, Calling AI Training Fair Use

In early September, the US Justice Department filed a roughly 20-page Statement of Interest with the Southern District of New York in the New York Times v. OpenAI copyright case, marking Washington's first formal intervention in the wave of AI-training copyright lawsuits. The filing squarely backs OpenAI, arguing that training generative AI/LLMs on copyrighted works is generally fair use and does not itself violate copyright law, and calling the plaintiffs' 'dilution' theory 'seriously flawed.' The NYT said it would strongly oppose the position. The filing is advisory and non-binding on the court.

Confirmed

• The DOJ stated the US has a significant interest in rejecting any argument that training LLMs on copyrighted texts violates copyright law, asserting that training AI models on copyrighted materials does not per se infringe.
• The filing describes such training as 'extraordinarily transformative' fair use, says the fourth fair-use factor supports this conclusion, and holds that 'dilution' is not a valid infringement claim.
• It warns that restricting AI training on copyrighted materials could 'impede creative and scientific progress,' and invokes national security, arguing that treating such training as infringement would harm US science, prosperity, and national security.
• Reuters, AP, Wired, The Guardian, The New York Times, and CTV News all confirmed the Trump administration/DOJ position backing OpenAI.
• Background: the NYT alleges OpenAI and Microsoft used millions of articles without permission to train chatbots; the case is before a Manhattan federal court, with multiple publishers joining similar suits since 2023. Per TheChuckTone citing AP, the stance could also significantly affect generative fields such as AI music.

Why it matters

• It is a rare, explicit government alignment in the AI copyright dispute that could shape the trajectory and precedents of the generative-AI copyright litigation wave.
• The NYT's dilution theory was directly challenged, and the NYT has vowed to fight back.
• Developer Chomba Bupe (via AlexTensor) warned that letting companies freely scrape works and have chatbots regurgitate them could flood the ecosystem with low-quality content at creators' expense.
• The filing is advisory and non-binding; the final say on fair use remains with the judge.

2026-09-02 ~ 2026-09-04 · 21 related posts

• US government tells OpenAI lawsuits: AI training is fair use, dilution theory 'deeply flawed' — Formal_Drop526 · 2026-09-02
• [source] DOJ tells SDNY that training gen AI on copyrighted works is lawful fair use — tszzl · 2026-09-02
• DoJ sides with OpenAI vs NYT: training LLMs on copyrighted works isn't infringement — Scobleizer · 2026-09-03
• Trump Admin Files Brief Backing OpenAI Against NYT, Calling LLM Training Fair Use — zephyr_z9 · 2026-09-03
• DOJ formally backs OpenAI: training LLMs on copyrighted text should be fair use — rohanpaul_ai · 2026-09-03
• US government formally backs OpenAI's fair use claim in NYT copyright case — rohanpaul_ai · 2026-09-03
• US government backs OpenAI in NYT copyright case: training is not infringement — Charuru · 2026-09-03
• NYT: Justice Department backs OpenAI, citing national security stakes in copyright case — nordicinst · 2026-09-03
• Trump administration sides with OpenAI in New York Times copyright lawsuit — wiredmagazine · 2026-09-03
• DOJ sides with OpenAI in copyright battle, warns restricting AI training could thwart progress — Polymarket · 2026-09-03
• DoJ sides with OpenAI on copyright; dev warns 'shortsighted' win will dry up training data — AlexTensor · 2026-09-03
• DOJ files 'historic' letter backing OpenAI in NYT copyright lawsuit — GaryMarcus · 2026-09-03
• DOJ Backs Argument That AI Training on Copyrighted Material Is Fair Use — TheChuckTone · 2026-09-04
• US government defends OpenAI's training on copyrighted content in NYT lawsuit — emmanuelvivier · 2026-09-04
• US government sides with OpenAI in NYT case, defending AI training on copyrighted content — emmanuelvivier · 2026-09-04

6 near-duplicate retellings: rohanpaul_ai · ChrisGPT · nordicinst · Ok_Homework_1859 · nordicinst · nordicinst