The New York Times has spent more than $20 million pursuing its copyright lawsuit against OpenAI and Microsoft, and publisher A.G. Sulzberger says the paper has no plans to stop. The case, filed in 2023, alleges the two companies illegally used Times journalism to train AI models, and nearly three years in, motions over discovery requests are still being filed. Sulzberger, who has held the publisher role for roughly a decade, framed the spend as the price of defending journalism industry-wide.
The Times can afford the fight. The paper recently surpassed 13 million subscribers, and its newsroom is one of the best-resourced in the business. That balance sheet is doing double duty: it is funding a landmark AI copyright case that will help set the legal frame for how frontier model builders source training data, while also underwriting a separate legal battle with the Trump administration over press access.
“It's the cost of defending journalism—at the Times and everywhere else.”— A.G. Sulzberger, Publisher of The New York Times
Sulzberger sat down with WIRED editor Katie Drummond in late July 2026 to discuss both fronts. The immediate flashpoint is a Justice Department subpoena campaign targeting Times reporters who covered the Qatari-donated jet intended for use as Air Force One, a gift Sulzberger noted is worth hundreds of millions of dollars. The subpoenas extended to third-party phone service providers, seeking records not just from journalists but from their family members.
Key facts
- 01The New York Times has spent more than $20 million pursuing its copyright case against OpenAI and Microsoft, filed in 2023.
- 02The Times has surpassed 13 million subscribers, giving publisher A.G. Sulzberger the financial base to sustain the fight.
- 03Sulzberger says the paper put reporters on the ground in more than 150 countries last year.
- 04The suit alleges OpenAI and Microsoft illegally used Times journalism to train AI models; discovery motions are still active.
- 05Sulzberger has held the publisher role for roughly a decade and sat for the interview in late July 2026.
The paper's reporting had established that on the jet's first overseas trip, to Turkey, the aircraft was deemed insufficiently equipped to respond to a threat, forcing the president to fly back on the original Air Force One. The administration initially denied the account, then it was confirmed. What Sulzberger called unusual was not the story upsetting a sitting president — every administration produces those — but the speed with which subpoenas were served at reporters' homes.
The AI case sits on a parallel track. The Times argues that OpenAI and Microsoft copied and ingested millions of its articles to train the models underpinning ChatGPT and Copilot, then built commercial products that compete with the paper for reader attention. OpenAI has argued its use is transformative and covered by fair-use doctrine. The legal outcome will shape the economics for every publisher whose archive has become model training fuel.
Sulzberger's argument to peer publishers is procedural rather than moral. Rights atrophy when unused. He pointed to the Associated Press suing the White House over the Gulf of America renaming access dispute as an example of a suit worth filing regardless of the win probability. He also criticized what he described as chilled behavior at some newsrooms, where reporters have told Times colleagues in Washington that editors are killing stories out of fear of retaliation and waiting for the Times to publish first.
“Your rights only hold if you seek to enforce them. You may not win every battle, but the surest way to lose your rights is to avert your eyes, drop your head, and try to avoid notice and just take what comes.”— A.G. Sulzberger, Publisher of The New York Times
The Times investment in original reporting is the counterweight Sulzberger keeps returning to. The paper had reporters on the ground in more than 150 countries last year — the kind of foreign bureau footprint that most of the industry has spent two decades cutting. It is also the exact category of expensive, high-friction journalism that generative AI products consume the most efficiently, summarizing and rerouting the traffic that would otherwise reach the publisher.
The closest recent parallel Sulzberger cited for the reporter subpoenas came during the first Trump administration, when some Times reporter records were secretly seized by the Justice Department, prompting reforms that were supposed to curtail exactly this kind of overreach. A similar raid at a Washington Post reporter's home followed months earlier. Sulzberger characterized the current environment as the biggest effort to go after the press from a White House since Woodrow Wilson, more than a century ago.
The counterweight to Sulzberger's position is real. OpenAI and Microsoft have signed licensing deals with the Associated Press, News Corp's Wall Street Journal parent, and other publishers, arguing that the market is settling toward negotiated rates rather than court-ordered damages. A Times courtroom win could unlock retroactive licensing bills across the industry, but a loss — or a narrow ruling on fair use — could formalize the current status quo, where scraped archives fuel commercial models with no compensation flowing back.
The Times case is the closest thing the AI industry has to a bellwether copyright trial, and its cost tells publishers something important: fighting a well-capitalized model provider through discovery alone runs into the tens of millions before a jury is ever seated. Most newsrooms cannot write that check. If the Times wins, the settlement math changes for every publisher negotiating with a foundation model lab. If it loses, the licensing deals already signed become the ceiling rather than the floor, and the papers that held out for a courtroom outcome will find themselves negotiating from a much weaker position.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
