We think the DOJ OpenAI copyright brief is good news for model makers and almost no news for you. Your risk sits in what you publish and where you sell, so stop reading the headlines and go read your own contract.
The short version
- The Justice Department backed OpenAI’s fair use defence on training, but a statement of interest doesn’t bind Judge Stein, and the brief itself says outputs are a separate question.
- That separate question is the one you live with. You publish what the tool writes, and you may sell into countries with no fair use doctrine.
- Our advice is to check which plan and indemnity you’re on, keep an edit trail, and don’t post long AI passages untouched.
The filing landed on 1 September, according to Bloomberg Law, in the consolidated New York case where the New York Times and other publishers are suing OpenAI and Microsoft. The publishers have pushed back, and a decision is some way off. Our earlier look at AI-generated video and copyright risk covers the same split between how a model learns and what a business publishes.
What does the DOJ OpenAI copyright brief say?
It says training a model on copyrighted writing is fair use, and it asks the court to reject arguments otherwise. It also says output questions should be judged separately and concedes that some output uses may not be transformative. The government isn’t a party to the case, and the judge isn’t bound by anything in it.
Per a legal summary of the 19-page filing, the government calls training “transformative, spectacularly so” and warns that heavy licensing requirements could entrench a few large technology firms. The same summary says the brief rejects a theory that AI output can harm a publisher’s market even without copying its text. Axios reports the filing also argues that outputs aren’t necessarily covered by the training argument. That last point is the one we’d underline twice.
Why doesn’t a training win protect a business that uses AI?
Because you’re standing on the other side of the transaction. You didn’t train the model. You publish what it writes, and you may sell into places where US fair use doesn’t apply. A ruling on training leaves both of those exposed, and most coverage skips straight past them.
Start with outputs. The brief itself separates them from training and says reconstructed and redistributed material may not be transformative. Picture your marketing lead pasting three paragraphs that closely track a publisher’s article onto your blog. A training ruling does nothing for that page.
Then jurisdiction. The Next Web notes that the EU has no fair use doctrine, only a closed list of exceptions, and that the AI Act’s Article 53 asks general-purpose model providers to respect rightsholders’ opt-outs. It also reports a Munich court ruling against OpenAI over memorized song lyrics, now under appeal. A US win wouldn’t travel. German courts are already testing AI liability, as our piece on the Munich ruling over AI disclaimers explains.
What should a 50-person firm do about it this week?
Run four checks and write down the answers. Which plan are you on, and what does its indemnity cover? Who edits AI text before it goes out? Do you sell into the EU? Do any of your contracts promise clients original work? None of it needs a lawyer to start, only a spreadsheet and an afternoon.
On the first check, OpenAI’s Copyright Shield, announced in November 2023, reportedly covers ChatGPT Enterprise and API customers for output claims, not free or Plus users. Proskauer’s summary doesn’t spell out the carve-outs, so read them in your current terms and skip any 2023 summary, ours included. If your team writes client work on a personal plan, you may have no indemnity at all. Who owns the work you produce is a separate matter, which our piece on AI-generated work and copyright takes up. For comparing vendor terms side by side, our ChatGPT, Claude and Gemini business comparison is a reasonable start.
Read the fine print before the headline
Three clauses matter more than any ruling. First, what the indemnity covers, whether outputs, inputs or both, and on which plans. Second, the carve-outs, which usually remove cover if you ignored a filter, fed in material you didn’t own, or edited the output in ways the vendor calls material. Third, who controls the defence and settlement. An indemnity you can’t use without the vendor’s consent is a weaker promise than it reads.
The contract side matters as much as the court side, and our look at the Musk v. OpenAI verdict and vendor contracts is a reminder to read the paper. Here’s a worked example. A 60-person agency posts an AI-drafted client article from a personal-tier account. A publisher alleges the piece reproduces its reporting. The agency’s contract promised the client original work. The model vendor’s indemnity, if one exists on that plan, is a separate contract the client never signed. The agency holds the risk. That’s an illustration, not a case, but each link in it can be checked against your own paperwork.
What would change our mind
We’d soften this if Judge Stein ruled on outputs as well as training and sided with the vendors on both. We’d also soften it if the big vendors extended their indemnities to every paid tier with no carve-outs. Until then, the training question is the easy half. Our read of the brief rests on public coverage and a published legal summary, which differ in emphasis, and a headline reference to national security isn’t something we lean on. Nothing here is legal advice, and we can’t say how the judge will rule.
The sceptic’s best case
The sceptic says we’re overcomplicating it. If the US government and probably the courts treat training as fair use, the cost of AI falls, vendors defend their customers, and small firms should get on with using the tools. That’s a fair read, and the brief does tilt the odds toward model makers on training.
Where we part ways is the step from “training is probably fine” to “I’m covered”. Those are different claims, and the second depends on a contract most owners haven’t opened since they clicked accept.
What to watch
- The 6 November 2026 reply briefs on the summary judgment motions, per the legal summary.
- Whether Judge Stein addresses outputs separately from training, as the DOJ asked.
- Any vendor that expands or narrows its customer indemnity after the filing.
- How the Munich appeal on song lyrics moves.
Frequently asked questions
Does the DOJ brief make AI training legal?
No. A statement of interest gives the court the government’s view and doesn’t bind it. Judge Sidney Stein will decide the fair use question, and the ruling could be appealed.
Am I covered if my team publishes AI-written content?
It depends on your plan and your vendor’s terms. OpenAI’s Copyright Shield reportedly covers ChatGPT Enterprise and API customers for output claims, but not free or Plus users. Read the carve-outs in your current agreement.
Does a US fair use ruling apply in Europe?
Not directly. The EU has no fair use doctrine and relies on a closed list of exceptions, so a US win on training would not carry over to sales or publishing there.
Written by Marcus Laporte, an AI editorial persona at AI Magazine Canada. This is analysis and opinion. Archive entry dated 7 September 2026, written and fact-checked on 8 October 2026. Sources are linked on the claims they support.
