• Tech Tech

New York Times, OpenAI, and Microsoft clash as copyright fight turns pivotal

If OpenAI and Microsoft win, publishers may have far less power to limit or profit from scraping, reusing, and monetizing their work.

The New York Times building.

Photo Credit: iStock

A major legal fight involving The New York Times, OpenAI, and Microsoft is entering a consequential phase, with the outcome likely to influence both the economics of the AI industry and the rules around using published content to build these systems.

Here's what to know

The copyright case the Times filed in 2023 has now reached the point where each side has laid out its arguments before District Judge Sidney Stein. According to Axios, he is expected to decide whether all or part of the dispute should proceed to trial in 2027.

The Times says OpenAI and Microsoft copied its reporting at large scale and used it to develop products that compete with its journalism. The two companies respond that copyright law permits training large language models on material that is publicly available, and they also argue that factual information is not itself covered by copyright.

OpenAI has relied on two California federal decisions — Kadrey v. Meta and Bartz v. Anthropic — where judges concluded that using copyrighted material to train AI models qualified as "fair use" because the training was transformative. Microsoft has advanced comparable reasoning in defending its Copilot product.

If the Times prevails, the revenue model supporting many AI products could face serious disruption.

If OpenAI and Microsoft win, publishers may have far less power to limit or profit from scraping, reusing, and monetizing their work.

More background

Observers well beyond New York are watching the lawsuit as one of the most important copyright tests yet for generative AI. Its eventual outcome could affect publishers, writers, artists, tech firms, and the growing number of people who use AI-generated responses.

The Times is not directly challenging the California decisions. Instead, it says those cases involved different facts because the plaintiffs there did not demonstrate the same type of market competition — namely, AI-generated outputs that function as substitutes for its journalism. The newspaper also argues that the New York court should follow Supreme Court and Second Circuit precedent.

The federal government has also entered the dispute. The Justice Department submitted a brief backing OpenAI and said that training large language models on millions of digital works creates a "transformative public benefit" that outweighs competitive harm to publishers.

This legal fight also connects to the energy grid. Building and running advanced AI systems requires power-hungry data centers and significant water use, which can strain local resources and raise energy costs. At the same time, AI can help utilities forecast demand, improve grid efficiency, and support cleaner energy systems. The case is part of a broader debate over how AI should grow — and at what cost.

What's being done?

The dispute is mainly playing out through court filings and rulings. Judge Stein's summary judgment decision could sharply reduce the issues in the case or clear the way for a closely watched trial, offering the next big indication of what AI companies may legally do with scraped material.

At the center of the dispute is a fundamental disagreement over how to characterize AI training and its outputs. OpenAI argues the practice can amount to "fair use," while the Times contends the resulting systems can operate as a "commercial substitute" for the journalism they rely on.

Where can I learn more?

The lawsuit is about copyright, but it also points to a broader question: what AI growth looks like on the ground.

• Across the U.S., backlash over AI growth is rising over jobs, water, electricity, and security.

• Big Tech's AI boom is outrunning the power grid as infrastructure spending races ahead of supply.

That mix of pressures helps explain why AI policy now reaches well beyond the courtroom. Rules for training models are colliding with the infrastructure, costs, and public buy-in the industry needs to keep expanding.

Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.

Cool Divider