New York Times Claims OpenAI and Microsoft Acknowledged AI Training as Theft

The legal conflict between The New York Times and the partnership of OpenAI and Microsoft has intensified following the release of new court documents. The newspaper alleges that leadership at both technology firms were fully aware that their methods for training artificial intelligence models involved the unauthorized use of copyrighted journalism. According to the filing, these companies proceeded with their data collection strategies despite internal recognition that such actions constituted a significant infringement on intellectual property.

Central to the latest allegations is the claim that millions of news articles were ingested to build commercial AI products without obtaining the necessary permissions. The New York Times characterizes this practice as an unprecedented theft of labor, arguing that the tech giants prioritized the development of their systems over the legal rights of content creators. The filing suggests that executives understood the potential for these AI tools to act as direct substitutes for traditional news outlets, thereby creating an existential threat to the publishing industry.

Internal communications cited in the documents indicate that staff members at OpenAI were cognizant of the limitations of their own mitigation efforts. Even as the companies implemented citation features to link back to original sources, internal admissions reportedly revealed that users rarely engaged with these links, undermining the argument that the AI tools were merely driving traffic to news sites. This evidence is being used by the plaintiffs to challenge the defense that their use of news content qualifies as transformative fair use.

Beyond the specific claims regarding The New York Times, the lawsuit has expanded to include other media organizations, such as Ziff Davis and the parent company of The Intercept. These publishers are seeking damages for the alleged misappropriation of their work. While Microsoft and OpenAI maintain that their activities are legally protected, the latest disclosures aim to demonstrate that the companies knowingly bypassed paywalls and stripped copyright notices to fuel their massive training datasets.

As the litigation continues in federal court, the outcome remains uncertain. The tech companies have previously dismissed the infringement claims as meritless, asserting that their technology provides a new, innovative service. However, the introduction of these internal documents marks a pivotal moment in the case, as the plaintiffs attempt to prove that the defendants were not merely negligent, but acted with full knowledge of the potential harm caused to the news ecosystem.