Internal Messages Expose OpenAI and Microsoft to the Media

Media groups, including The New York Times, are demanding billions from OpenAI and Microsoft. Their case outlines internal communications, data, and cost estimates that describe substitute services and disputed practices, while aiming to reject the fair use doctrine.

Costs and Competition: One Million Articles for $6,800

The plaintiffs argue that at OpenAI's API rates, producing one million 500-word articles costs about $6,800, without the involvement of journalists. They contest the notion that AI models do not compete with publishers and assert that OpenAI's products enable the mass generation of "pink slime" content—low-quality pseudo-news, often plagiarized. Internal data from Microsoft shows that click-through rates on Copilot are significantly lower than those on traditional Bing search. Microsoft refers to a "doom loop," a self-reinforcing cycle that, according to the company, would harm both model performance and the web as a whole.

Disputed Practices: Paywalls, Licenses, and Post-Lawsuit Filters

Systematic circumventions of paywalls and violations of terms of service are attributed to OpenAI, with a message where Brockman responds with "oh nice" to a hack targeting The New York Times' paywall. The plaintiffs also claim that a filter was designed immediately after the lawsuits were filed to remove outputs likely originating from their publications, not to protect copyright but to hinder evidence collection. They further assert that rather than compensating for their content, the defendants took it for free and exchanged it among themselves.

Fair Use Undermined by Internal Statements and Federal Position

The plaintiffs refer to internal communications and sworn depositions in which officials express doubts about the legitimacy of fair use and present chatbots as replacements for original journalism. Brent Hecht, Director of Applied Sciences at Microsoft, described training on protected content as "astonishing theft of unprecedented scale" and potentially "the largest theft of labor in human history." He also estimated that a successful fair use defense "would completely mock the idea of 'fair use'." Microsoft clarified that these remarks reflect an individual employee's perspective and do not constitute legal analysis. At OpenAI, Nick Turley mentioned an "existential threat" to publishers and "largely substitutive" products that are increasingly set to replace their offerings. In May 2025, the U.S. Copyright Office concluded that fair use cannot generally apply to the mass copying conducted by AI companies.

What the Plaintiffs Are Demanding and Company Reactions

A number of American media outlets, including The New York Times, are demanding billions from OpenAI and Microsoft due to alleged copyright violations. They seek acknowledgment of these infringements, rejection of the fair use defense, and that each article be considered an individual work for the assessment of damages. Microsoft clarified that internal criticisms regarding fair use represented a personal viewpoint and not a legal assessment. At the same time, Microsoft has entered into licensing agreements with other media groups, a move also followed by Amazon, Google, Meta, and Perplexity.