USA Today Co. and several newspapers it owns have sued OpenAI in a Manhattan federal court, accusing the artificial intelligence company of copying hundreds of thousands of news articles and other copyrighted materials without permission to train the technology behind ChatGPT.
The lawsuit, filed on Thursday in the US District Court for the Southern District of New York, seeks damages in excess of $250 million and a court order prohibiting OpenAI from continuing the alleged infringement. The case adds to mounting legal pressure on AI developers over their use of copyrighted journalism, books, and other creative works to build and improve large language models.
USA Today alleges that OpenAI used its publications’ content without authorization during the training process for its AI systems. The complaint places the dispute within a broader conflict over whether technology companies can use large volumes of copyrighted material to develop commercial AI products without obtaining licenses or compensating the original rights holders.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
The lawsuit comes as publishers and other copyright owners pursue legal action against major technology companies, arguing that the commercial value of their work is being used to develop competing products without adequate permission or payment. Technology companies, in turn, have argued that training AI models on copyrighted material constitutes fair use because the systems transform the material into something new.
The dispute has become one of the most consequential legal questions facing the generative AI industry, with potential implications for how companies obtain training data, negotiate licensing agreements and distribute the economic benefits of AI-generated products.
USA Today’s complaint is one of dozens of high-stakes lawsuits filed by copyright owners against companies including OpenAI, Anthropic and Meta Platforms over their alleged use of protected material in AI training.
The cases challenge a central feature of the industry’s development. Large language models are trained on extensive datasets containing text and other information, enabling them to generate responses, summarize material and perform a range of language-based tasks. Publishers have argued that when this process involves copyrighted articles, companies should not be able to use the material commercially without permission.
For news organizations, the dispute extends beyond the question of whether articles can be included in training datasets. Publishers also face concerns about how AI products may change the relationship between audiences and the original sources of information. Chatbots can answer questions and summarize subjects that users might otherwise research through news websites, raising questions about traffic, advertising revenue and the long-term economics of journalism.
The legal challenge is therefore about both intellectual property rights and the distribution of value in the AI economy. Publishers invest in reporting, editing and maintaining news operations, while AI developers seek access to extensive bodies of text to build systems that can compete for users’ attention. The lawsuits are testing whether existing copyright rules provide sufficient protection for publishers in that environment.
The financial stakes are significant. USA Today’s demand for damages exceeding $250 million indicates the scale of compensation it is seeking, although the amount requested in a complaint does not establish what a court may ultimately award. Its request for an injunction also seeks to restrict the alleged infringement, potentially placing limits on OpenAI’s use of the material at issue.
OpenAI Faces Another Legal Front As AI Companies Defend Fair Use
OpenAI and its largest financial backer, Microsoft, are already facing similar claims in Manhattan federal court from prominent authors and news organizations, including The New York Times.
The overlapping cases point to a widening confrontation between established content businesses and technology companies seeking to commercialize generative AI. Publishers are pursuing legal remedies through the courts, while AI developers have relied on fair-use arguments to defend their ability to train models using existing material.
At the center of the disagreement is whether the use of copyrighted content to train an AI model should be treated as a permissible transformation of that material or as an unauthorized commercial use that requires consent and compensation. The answer could shape the cost and availability of training data, particularly for companies whose products depend on large collections of professionally produced text.
A ruling that imposes significant restrictions on training practices could increase pressure on AI developers to negotiate licenses, establish clearer permissions, or modify how they source and process content. More licensing agreements could provide publishers with an additional source of revenue, but the commercial terms would depend on negotiations, the value of individual content libraries, and the legal outcomes of the disputes.
Conversely, a broad acceptance of technology companies’ fair-use arguments could preserve greater flexibility for AI developers while leaving publishers with fewer options to demand payment for material used in training. The precise implications would depend on the scope of any ruling and the facts of each case.
A coalition of artificial intelligence industry groups and author advocacy organizations, last year, urged a federal appeals court to block what they describe as the largest copyright class action ever certified, warning that the case could financially devastate not only Anthropic, the AI company at its center, but the entire generative AI industry.