Seattle Times and Newsday sue OpenAI and Microsoft, demanding destruction of AI models

The Seattle Times and Newsday have filed a lawsuit against OpenAI and Microsoft, alleging that the tech giants infringed their copyrights by using their journalism as training data for artificial intelligence models without authorization. The publishers claim that not only was the data used improperly, but the resulting AI systems often reproduce specific passages from their reporting in response to user queries. In a significant escalation of remedies sought in AI copyright cases, the plaintiffs are asking the court to order the destruction of any copies of their works held by the companies, including the training datasets and the AI models themselves that incorporate that content.

The Legal Demands

According to reporting by The Verge, the suit mirrors similar legal actions previously filed by high-profile media entities such as The New York Times, Ziff Davis, Merriam-Webster, and Encyclopedia Britannica. However, the specific request for the destruction of the models distinguishes this filing from earlier iterations of the dispute. The Seattle Times and Newsday argue that the unauthorized use of their content violates their exclusive rights to reproduce and distribute their work. By naming Microsoft as a co-defendant, the publishers are targeting the commercial distribution channel of the technology, specifically Microsoft Copilot, which is built on OpenAI’s underlying models. OpenAI and Microsoft did not immediately respond to requests for comment regarding the new allegations.

Context in the Broader Litigation Landscape

This lawsuit is part of a rapidly expanding legal front against generative AI companies. The publishers are joining a coalition of nearly 400 local newspapers that have recently sued OpenAI and Microsoft. That broader group of plaintiffs argues a different but related economic harm: that the chatbots reduce the need for users to visit their websites for news and answers, thereby directly costing them valuable subscription revenue. While the local newspaper coalition focuses on lost traffic and revenue, the Seattle Times and Newsday suit focuses more heavily on the intellectual property rights of the source material itself, seeking to remove the infringing data from the models entirely. This dual-pronged approach—targeting both the direct infringement of the data and the downstream economic impact on the industry—suggests a coordinated strategy among media organizations to challenge the foundational training practices of major AI labs.

Implications for Developers and AI Practitioners

For developers and companies building on large language models, this litigation highlights the growing legal risks associated with the provenance of training data. While the immediate impact on open-source or proprietary model development workflows may be limited until court rulings are issued, the demand for model destruction sets a high bar for potential outcomes. If courts were to mandate the deletion of models trained on copyrighted material, it could force a re-evaluation of how AI companies document and manage their training data pipelines. Developers should be aware that the legal environment for AI training is becoming increasingly adversarial, with plaintiffs seeking not just monetary damages but structural changes to how models are built and deployed. This could lead to a more fragmented landscape where models are trained on strictly licensed or public domain data, potentially affecting the breadth and depth of knowledge available in AI systems.

What to Watch

• Court Rulings on Model Destruction: Whether judges accept the argument that AI models can be "destroyed" or if they view the models as transformative works that cannot be unwound.
• Microsoft’s Response: How Microsoft defends its role as a distributor and developer of Copilot, and whether it settles to avoid setting a precedent for corporate liability in AI training.
• Expansion of Plaintiffs: Whether other major news outlets join the Seattle Times and Newsday in seeking similar remedies, potentially creating a class action that could overwhelm the defendants’ resources.
• Impact on Training Data Practices: If the lawsuit succeeds, it may force AI companies to implement more rigorous data provenance tracking and licensing agreements, which could increase the cost of training new models.