Superpower Daily

Superpower Daily

Your reading space

Sign in
LlamaIndex Begins Stainless Exit, Seeking to Preserve SDK Inter...Apple Releases macOS 27 With a Siri That Can Read Your ScreenAudit Finds No SWE-bench Pro Leaderboard Score Has a Complete P...Salesforce Connects Gemini Enterprise to Its Data Without Custo...Emergence Finds AI Agents Invent Opaque Shared Dialects in Coop...Meta Launches Meta One Subscriptions for More AI Across Its AppsLlamaIndex Begins Stainless Exit, Seeking to Preserve SDK Inter...Apple Releases macOS 27 With a Siri That Can Read Your ScreenAudit Finds No SWE-bench Pro Leaderboard Score Has a Complete P...Salesforce Connects Gemini Enterprise to Its Data Without Custo...Emergence Finds AI Agents Invent Opaque Shared Dialects in Coop...Meta Launches Meta One Subscriptions for More AI Across Its Apps
Swipe

OpenAI and Publishers Ask Court to Decide AI Training Fair Use

The next phase of publisher copyright cases will test whether model training transforms news content or competes with the businesses that produce it.

By 2 min read
OpenAI and Publishers Ask Court to Decide AI Training Fair Use
OpenAI and Publishers Ask Court to Decide AI Training Fair Use

Listen to this story

The audio brief

About 1:42
0:001:42
Read transcript
A federal court is being asked to decide a threshold question that could shape the future of AI and journalism: is training a model on copyrighted news reporting fair use? The request comes in consolidated lawsuits involving The New York Times and Ziff Davis, CNET’s owner, with separate claims against OpenAI and Microsoft. After extensive discovery, the judge could rule on those copyright issues before sending the cases back to their original districts for possible trials. OpenAI says training on material gathered from the internet is transformative. Its lawyers emphasize that news contains a large amount of factual information, and argue that model training does not create legally recognizable market harm. Publishers describe the same practice differently. They say it has reduced traffic to their websites and helped flood the information market with AI-generated “pink slime”—cheap, low-quality content that competes with original reporting. A second fight concerns access. OpenAI argues that publishers effectively allowed scraping when they failed to block its crawlers through robots.txt, a file that tells automated systems what they may access. Many publishers now block OpenAI and other AI companies, but whether earlier inaction amounted to an implicit license is unsettled. The Trump administration says licensing barriers could burden AI firms. Ziff Davis CEO Vivek Shah disputes that, comparing roughly twenty billion dollars in possible annual publisher payments with OpenAI’s projected seven-hundred-fifty-billion-dollar computing spend by 2030. The court’s central question remains: does training transform journalism, or undermine the market that produces it?

Story brief

3 key points

OpenAI and publishers have asked a federal court to resolve whether using copyrighted journalism to train AI models is fair use, potentially establishing rules before the cases proceed toward trial. OpenAI says the practice is transformative, relies heavily on factual material, and does not create legally recognizable market harm; publishers point to lost traffic and AI-generated “pink slime” competition. The...

  1. 01

    The consolidated cases involve The New York Times and Ziff Davis, CNET’s owner, alongside separate claims against OpenAI and Microsoft.

  2. 02

    The court may decide threshold copyright questions before sending the lawsuits back to their original districts for possible trials.

  3. 03

    Many publishers now block OpenAI and other AI crawlers, but the legal effect of earlier inaction remains unsettled.

The publisher lawsuits against OpenAI are now poised for a legal ruling that could shape how AI companies use copyrighted reporting. The parties have filed briefs asking a federal court whether training models on published work qualifies as fair use under U.S. copyright law.

The cases include claims from The New York Times and Ziff Davis, CNET’s owner. They are advancing after a lengthy discovery process in which separate lawsuits against OpenAI and Microsoft were consolidated before one judge. The court could decide key questions before the cases return to their original districts for potential trials.

OpenAI’s defense also turns on how publishers made their sites available. The company says publishers implicitly licensed access when they did not initially block its web scrapers through robots.txt or similar tools. Robots.txt is a website file that tells automated crawlers what they may access.

Many publishers now block bots from OpenAI and other AI firms. OpenAI presents those controls as an available opt-out route; the question before the court is whether the earlier absence of such blocks carries the legal effect the company claims.

Licensing is part of the dispute

The Trump administration has submitted a statement of interest arguing that licensing barriers could burden AI companies. Ziff Davis Chief Executive Vivek Shah disagreed with the premise that licensing would be prohibitively costly, comparing roughly $20 billion in annual payments to news publishers with OpenAI’s stated expectation of spending $750 billion on computing infrastructure by 2030.

Shah called a comparable licensing bill a “rounding error” for AI firms and said greater use of access barriers and paywalls would make information harder to reach. That is Shah’s argument, not a licensing proposal before the court.

The briefs do not resolve the dispute. They put the central disagreement squarely before the court: whether models make a legally distinct use of published work, or whether that use harms the market for the journalism on which they were trained.

Sources

  1. cnet.comPublishers Argue AI Firms Should Pay for Content as Copyright Case Advances - CNET

Loading discussion...

Related coverage

Continue reading