DOJ's 20-page brief backs OpenAI, argues AI training on copyrighted work is fair use
A filing with no binding force still moves the cost of walking away, and every publisher currently quoting a number to a lab is quoting it to a counterparty whose government has said in court that permission may not be required.
The Investor · Invest desk
What happened
- The Justice Department filed a brief on September 1, 2026 arguing that training AI models on copyrighted work is fair use, and warning that a licence requirement would hand the biggest firms a near-monopoly on model building.
- The government's economic case is that paying for training data would raise entry hurdles, strengthen incumbent tech firms, and route much of the resulting profit to publishers holding the largest archives.
- Reuters reports it as the first time the US government has entered the wave of AI copyright litigation, in a twenty-page filing that carries no binding legal authority of its own.
- OpenAI and The New York Times file summary-judgment motions on Friday, days after the government's intervention reached the Manhattan docket.
Why it matters
- What it limits constraint A publisher's negotiating floor is the cost of litigating rather than licensing, and the government has pushed that floor down before either side exchanges a number.
- Who has to decide decision Any lab facing a renewal has a reason to wait for the Friday motions to resolve, and any publisher has a matching reason to close before they do.
- What doesn't add up contradiction The Justice Department's entry-barrier theory and the Copyright Office's source-and-market test are both official US positions, so counsel on either side of a licensing table can now cite Washington.
- Who's exposed exposure Labs that already bought archive access are holding an asset whose permission component their own government has argued was never required, while the clauses it left undefended cover outputs and provenance.
Next story · Invest · Sep 21
A single Times call-spread buyer risks $407,000 on an October copyright rulingConfirmed4 publishers
A licence is worth the cost of not having one, which is why a document carrying no binding legal authority [8] can still move a price. The Justice Department did not change copyright law on Tuesday. In twenty pages it told a Manhattan federal court that the United States has a strong interest in the court rejecting any argument that training large language models on copyrighted texts violates copyright law [3][6], and it called that training extraordinarily transformative, with benefits that far outweigh any competitive harm [4].
The scale gap between the two sides is stark. Goldman Sachs Research projects more than a trillion dollars of AI-related investment globally in 2026, of which $581bn lands in the United States [19], so the domestic share the brief invokes is at most about 58 cents of every global dollar [26]. The New York Times, which sued OpenAI and Microsoft in 2023 [5], has spent more than $28m on AI-related litigation according to its regulatory filings [21], which is one dollar of legal spend for every 20,750 dollars of American AI capex this year [25]. That disparity does not settle who is right; it settles who can afford to keep going, and it is the same point the government makes about small developers facing licence fees [17], read from the other end of the table.
The two concentration stories are the interesting part. Washington says mandatory licensing would raise entry hurdles, bolster incumbent tech firms and transfer much of the profit to publishers with the largest archives [17]; the Times says the administration is siding with a handful of trillion-dollar AI companies at the expense of creators whose work they stole [12]. The OECD's reading, that access to computing power, data and expertise is already consolidated in favour of established firms [18], sits underneath both. The choice on offer is between a concentration priced in licence fees and one priced in GPUs, and the filing picks the second without saying so.
What the brief declines to defend is where the residual value sits. It separates training from outputs that reproduce copyrighted material, and from how the data was acquired [9], which is exactly the seam the California cases split on, since a Copyright Alliance reading of Bartz v. Anthropic and Kadrey v. Meta has both courts calling training transformative while reaching different conclusions on piracy and market harm [10]. The US Copyright Office holds that some training uses qualify and some do not, depending on source, purpose and market effect [11]. What a publisher still sells is clean provenance and indemnity, the parts of the deal the brief left standing.
This is probably wrong, but the deals signed over the next two quarters look to me like data-supply and indemnity contracts at lower headline numbers, because the permission component has been publicly marked down. The counter-thesis is simple: if the court follows the Copyright Office's factor-by-factor line and treats output substitution as dispositive, damages exposure becomes real and licence prices rise. A third path is that judges treat an executive-branch statement of interest in a private suit as the non-binding paper it is [8], the sanctions fight over allegedly destroyed evidence prices on its own [22], and the only durable effect is diplomatic, since Commerce Secretary Howard Lutnick spent Wednesday urging G20 officials in North Carolina to adopt fair use while devising ways to protect artists [23] rather than building a licensing framework, and the EU already runs the other model, with text-and-data-mining exceptions that rights holders may reserve against [24].
I would drop the read if a licence signed after Friday clears at a premium to comparable pre-filing deals, or if the opinion adopts the transformative framing and negotiations continue at unchanged numbers, which would mean they were never about litigation risk.
What to watch
- Whether Friday's summary-judgment opinion engages the statement of interest or sets it aside as non-binding paper.
- The sanctions motion brought by the Times and sixteen other outlets, which prices independently of the fair-use question.
- Whether any G20 member follows Lutnick's appeal or stays with the EU model, where rights holders can reserve works from mining.
Information, not investment advice. This story is general information about companies and markets, drawn from cited reporting. It is not a recommendation to buy, sell or hold any security; check the cited sources before acting.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence72
- Adoption
- Not scored
- Hype gap+20
- Incentives65
- Confidence68
Perspective Coverage
5 publishers- Builder
- Builder 26%
- Operator
- Operator 40%
- Investor
- Investor 34%
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
The US Justice Department filed a court brief on September 1, 2026, arguing that training AI on copyrighted work is fair use, and warning that requiring licences would hand the biggest tech firms a near-monopoly on model building.
- [2]
The Justice Department filed a statement of interest in the consolidated copyright lawsuit against OpenAI, telling the court that if developers were made to license training text, only wealthy corporations could continue making cutting-edge AI.
- [3]
The Justice Department filed a 20-page brief in Manhattan federal court on Tuesday siding with OpenAI in its copyright dispute with The New York Times, arguing that training LLMs on copyrighted material generally constitutes fair use.
- [4]
The Justice Department leaned on the first fair-use factor, describing AI training as "extraordinarily transformative" and arguing that AI's benefits "far outweigh any competitive harm."
- [5]
The New York Times sued OpenAI and Microsoft in 2023, arguing the company was effectively stealing billions of dollars' worth of its journalists' work, in some cases reproducing Times material verbatim.
- [6]
The brief said "The United States has a strong interest in this court rejecting any argument that training LLMs on copyrighted texts violates copyright law," citing scientific advancement and national security.
- [7]
The Justice Department told the court that siding with The New York Times and other publishers would thwart "creative and scientific progress while hindering American prosperity and economic mobility."
- [8]
According to Reuters, the filing marks the first instance of the US government stepping into the growing wave of AI copyright litigation, and the filing carries no binding legal authority.
- [9]
The filing does not give AI developers total protection: it differentiates model training from outputs that reproduce copyrighted material, and from issues regarding how the training data was obtained.
- [10]
A Copyright Alliance analysis of Bartz v. Anthropic and Kadrey v. Meta states that both courts ruled training was transformative but reached different conclusions on piracy and market harm.
- [11]
The US Copyright Office took a more nuanced stance, saying some uses of training AI can be covered by fair use while others may not, depending on the source of the material, the purpose of its usage, and the market impact.
- [12]
The New York Times said the administration "is siding with a handful of trillion-dollar AI companies at the expense of the countless American creators whose work they stole."
- [13]
Times spokesperson Graham James said "Both AI and creators can thrive - AI companies simply need to pay fairly for the content that makes their products possible, as copyright law requires."
- [14]
Associate Attorney General Stanley Woodward Jr. called the filing a "historic statement of interest" linking the administration's copyright position to keeping the US position in AI.
- [15]
Copyright holders have sued OpenAI, Anthropic and Meta Platforms over the use of their work in AI training.
- [16]
The government's statement arrives just before OpenAI and The New York Times file summary-judgment motions on Friday.
- [17]
The government argued that mandatory licensing could create huge hurdles to entry, bolster incumbent tech firms, and transfer much of the resulting profits to existing publishers with the largest archives, marginalising smaller developers.
- [18]
An OECD report says current AI market access to computing power, data and expertise is already consolidated and will continue to benefit companies already well established in the field.
- [19]
Goldman Sachs Research projects more than $1 trillion in global AI-related investment in 2026, including $581 billion in the United States.
- [20]
Authors Guild CEO Mary Rasenberger told WIRED she was "extremely disappointed" and called the filing "replete with faulty arguments and a gross misunderstanding of the fair use doctrine and copyright law."
- [21]
The New York Times has spent more than $28 million on AI-related litigation, according to regulatory filings.
- [22]
The Times and 16 other media organizations sought sanctions against OpenAI earlier this year, alleging it withheld and destroyed evidence, including by rendering billions of ChatGPT exchanges inaccessible; OpenAI denied the allegations.
- [23]
US Commerce Secretary Howard Lutnick addressed G20 officials gathered in North Carolina on Wednesday, urging nations to adopt fair use principles that permit AI companies to use creators' work for training while devising ways to "protect artists."
- [24]
The European Union's Digital Single Market Directive provides explicit text-and-data-mining exceptions, although rights holders can reserve their works from some forms of mining.
- [25]
The Times' AI litigation spend of more than $28m equals one dollar for every 20,750 dollars of projected 2026 US AI-related investment.
- [26]
The projected $581bn of US AI-related investment in 2026 is at most about 58 percent of the projected global total, which the source gives as more than $1 trillion.
Sources
4 independent publishers whose own reporting we read for this story.
- apnews.comTrump administration backs OpenAI in New York Times' copyright case | AP News
1 article · September 3, 2026
- cryptopolitan.comUS in support of OpenAI’s fair-use claim and warns of AI licensing oligopoly
1 article · September 2, 2026
- livemint.comTrump administration backs OpenAI in New York Times copyright case over training of chatbots | Mint
1 article · September 2, 2026
- pivotnews.aiJustice Department backs AI companies on fair use in New York Times case
1 article · September 3, 2026
- qz.comTrump administration backs OpenAI in NYT copyright lawsuit
1 article · September 2, 2026
Topics and entities
Entities
- OpenAI
- US Copyright Office
- Microsoft
- Mary Rasenberger
- Goldman Sachs
- Stanley Woodward
- U.S. Department of Justice
- Howard Lutnick
- The New York Times
- Meta Platforms
- Sidney H. Stein
- Shira Perlmutter
- Anthropic
- EU Digital Single Market Directive