DOJ Supports OpenAI
Justice Department Backs OpenAI in Defense of Fair Use for AI Training
Anonymous
Key Points
1
The DOJ argues that training LLMs on copyrighted text does not copy the source material; instead, the models use the text to develop general reasoning and language abilities, a process described as extraordinarily transformative under fair use, with researchers already achieving major breakthroughs as a result.
2
The government warns that licensing costs for training data could create an oligopoly on model training, advantaging the largest tech firms and potentially stifling competition and innovation in AI.
3
EU law lacks a fair use doctrine; it relies on a text and data mining exception with an opt-out, and the AI Act requires providers to identify and respect reservations, attaching obligations to the model itself rather than to the training run.
4
The DOJ filing is a statement of interest, not a binding ruling since the department is not a party to the case, meaning its views guide the court but do not decide the outcome.
5
Publishers criticized the administration’s stance as siding with large AI companies at the expense of creators, noting President Donald Trump’s push for AI leadership and a broader policy backdrop that shapes the debate over rights and innovation.
The Trump administration filed a 20-page brief supporting OpenAI in its copyright battle with The New York Times, arguing that training AI models on copyrighted news articles is fair use. DOJ claims the practice is transformative and vital for national security, economic growth, and scientific progress. The government warns that requiring licenses for training data would create an AI monopoly favoring the largest tech firms.
The move pits the Trump administration's push for American AI dominance against publishers who say they deserve payment when their work trains AI systems. The New York Times and other news outlets contend that OpenAI profits from their reporting without compensation, threatening newsroom livelihoods. The case reflects a high-stakes clash over who controls the future of artificial intelligence.
Why the Government Says Training Is Fair Use
DOJ argues that feeding copyrighted articles into AI models does not actually copy the source material. Instead, the text teaches models general language and reasoning skills—a process the department calls "extraordinarily transformative." The government frames AI training as analogous to human learning: students read books to gain knowledge, not to steal the text itself.
Under fair use law, transformative uses that change the purpose of copyrighted work may be legal without permission. DOJ contends that OpenAI's models use articles to build reasoning abilities, not to republish news. Researchers have already achieved major breakthroughs using this approach, the government argues, benefiting scientific progress and national competitiveness.
The Monopoly Risk of Licensing Requirements
DOJ warns that forcing companies to license training data would crush smaller AI competitors and entrench the largest tech firms. If startups must pay for every word used in training, only well-funded giants like OpenAI, Google, and Meta could afford to build models. Smaller innovators would be priced out of the market.
The government frames licensing requirements as a barrier to American innovation in a field where the US must compete globally. National security depends on maintaining AI leadership, DOJ argues, and a licensing regime would slow breakthroughs and hand advantage to international competitors. The brief emphasizes economic mobility: open AI access creates opportunity for entrepreneurs and researchers worldwide.
Publishers Push Back Against the Administration
The New York Times and other news publishers say the government is backing tech giants at creators' expense. ETV Bharat reports that critics argue newsrooms lose revenue when their work trains AI without licensing deals. Some publishers worry that if AI companies avoid paying for content, news organizations cannot fund investigative journalism.
The administration's stance reflects President Trump's broader push for AI dominance under his second term. Critics see the DOJ filing as siding with Silicon Valley over creators, potentially weakening copyright protections that have long protected artists, writers, and journalists. The outcome could reshape how news organizations and other creators are compensated in the age of AI.
How Europe Takes a Different Approach
The European Union does not recognize fair use the way US law does. Arcamax notes that EU law includes a text and data mining exception that lets companies train on copyrighted material—but publishers can opt out. The EU's AI Act requires companies to respect those reservations and identify which creators have claimed rights.
This opt-out system attaches obligations to the AI model itself rather than to each training session, creating a different balance between innovation and creator rights. The EU's stricter approach contrasts sharply with DOJ's position that US law should favor AI development. The diverging strategies highlight a global race for AI leadership where copyright rules shape competitive advantage.
Sentiment
Mostly NegativeNEGATIVENEUTRALPOSITIVE
Publishers
16
Articles
81
Reach
97
A
AnonymousAuthor
