New York Times Publisher Says AI and Government Pressure Are Testing Journalism’s Future

The New York Times is fighting a two-front battle: a high-stakes copyright case against OpenAI and Microsoft over AI training data, and what its publisher, A.G. Sulzberger, describes as a renewed campaign by the Trump administration to intimidate journalists. Sulzberger says both threats could reshape how journalism is funded, produced and protected in the AI era.

In a wide-ranging interview recorded in late July at WIRED’s offices, Sulzberger argued that major newsrooms must defend press freedom more aggressively while also being careful about how they adopt artificial intelligence. He said the Times views AI primarily as a reporting tool, not a replacement for human editors and reporters, even as the company’s subscriber business remains strong.

The conversation comes at a moment when the newspaper is reporting more than 13 million subscribers, handling legal discovery in its lawsuit against OpenAI and Microsoft, and responding to subpoenas tied to its coverage of a Qatar-gifted jet used by the Trump administration. For Sulzberger, the message was clear: journalism is being squeezed by political retaliation and by technology companies built on the content it produces.

Why A.G. Sulzberger says this moment is different

Sulzberger believes American journalism is facing an unusual convergence of pressure. One force is political: the Trump administration’s efforts to investigate, subpoena and in some cases publicly challenge major newsrooms. The other is technological: AI companies using vast quantities of published material to train systems that can summarize, remix and increasingly replace traditional search and information habits.

He said the current climate feels more threatening than a normal dispute between a president and the press because, in his view, the government is not simply criticizing coverage but taking steps that appear designed to chill it.

“This is the biggest effort to go after the press that we’ve seen from a White House probably since Woodrow Wilson,” Sulzberger said, arguing that the pressure on reporters is unlike anything he has seen in his decade as publisher.

That comparison is striking, but the point he was making was not historical precision so much as severity: he sees the current environment as one in which legal, regulatory and rhetorical pressure are increasingly used together.

What sparked the latest clash with the Trump administration?

The latest flashpoint was reporting by Times journalists on a luxury aircraft gifted by Qatar and used by the Trump administration as a presidential transport. According to Sulzberger, the paper’s reporters found evidence that the plane was not ready for secure presidential use on an early foreign trip, and that the president had initially denied those concerns.

He said the Justice Department’s response — subpoenas directed at several Times journalists and demands for communications records from third-party phone providers, including records connected to family members — crossed a serious line. In his view, it was not just a routine legal maneuver but an attempt to intimidate the reporters involved.

He also said the subpoenas were served very quickly and in a way that departed from longstanding norms and official guidelines meant to protect journalists from overreach.

How unusual are subpoenas aimed at reporters?

They are unusual enough to alarm newsroom leaders, Sulzberger said. He pointed to the first Trump administration, when Justice Department actions involving reporter records prompted reforms designed to limit such abuses. He also referenced other recent cases in which journalists’ homes or devices were targeted.

His broader warning is that even if these episodes are not completely unprecedented, they mark a troubling escalation. When police power and legal process are aimed at reporters’ private communications, he said, the effect can be to make newsrooms think twice before pursuing sensitive stories.

How is the Times responding inside the newsroom?

Sulzberger said the Times is trying to send a message of support to reporters, especially those doing work that can draw political retaliation. He said he personally called journalists who were served subpoenas and told them the paper would back them.

The response inside the newsroom, he added, was not retreat but resolve. In his telling, the reporters were determined to keep going and not let the legal pressure derail the story.

According to Sulzberger, the internal message was simple: the company would “have your back” and do what was needed to support the journalists targeted by the subpoenas.

That posture reflects a larger newsroom strategy. The Times, he said, wants staff to understand that the institution will defend reporting that is done carefully and in the public interest, even if that reporting angers powerful figures.

What does Sulzberger think other newsrooms should do?

He argues that media organizations cannot preserve their rights by avoiding conflict. In his view, legal protections for the press matter only if publishers are willing to use them when challenged.

He also said he does not believe the industry has broadly surrendered to pressure from the Trump administration. Instead, he sees a mixed landscape: many outlets are holding firm, while some are becoming more cautious, deciding not to pursue stories they fear might trigger retaliation.

That caution, he warned, can become a form of self-censorship. Once reporters and editors begin assuming that only the Times or another large institution will take the risk, the result is a narrower public conversation.

What role does AI already play in The New York Times newsroom?

AI is being used at the Times as a helper rather than an author. Sulzberger said the company is wary of turning over content creation to software, especially after seeing examples of news organizations publish AI-generated material that contained errors or invented details.

For him, the lesson is simple: the human newsroom remains accountable for everything that goes into print or onto a website. AI may assist with tasks, but it cannot own the judgment.

Sulzberger said AI should be treated as a tool, not a substitute for editorial responsibility, adding that newsroom leaders must catch and correct errors rather than assume a machine will get them right.

He described the Times’ approach as deliberately slow and experimental. The paper, he said, prefers to test systems quietly, study the results and learn where AI can help without compromising standards.

Where does AI help journalists most?

According to Sulzberger, AI is especially useful when a newsroom is dealing with large data sets, where the challenge is not writing copy but finding patterns, anomalies and leads.

He pointed to reporting that involved scanning thousands or millions of items, such as data-heavy investigations and visual analysis. In those cases, AI can act like a filter that reduces the volume of material reporters need to review.

That does not mean the machine makes the conclusion. Instead, the journalist still has to inspect the output, verify the evidence and decide what is newsworthy.

• AI can surface patterns in campaign finance and other large databases.
• It can narrow huge pools of images or documents to a manageable review set.
• It can speed up the earliest phases of reporting without replacing editorial judgment.
• It cannot be allowed to publish independently on behalf of a newsroom.

Why Sulzberger says AI has an “original sin”

For Sulzberger, the central problem with the AI boom is not just technical. It is economic and ethical. He argued that the industry’s foundations rest on four inputs: the people writing the code, the computing infrastructure, the electricity required to power it and the body of content that makes the systems useful.

In his view, the first three inputs are being paid for at extraordinary levels. Top engineers are commanding massive compensation packages, while companies are raising enormous sums of capital and building costly data-center and chip infrastructure. But the fourth input — books, journalism, music, films and other creative works — has often been taken without consent or payment, he said.

That, he believes, is the core injustice.

His argument is not only about the Times. It is about the broader ecosystem of content producers whose work underlies AI products but whose business models may be weakened if that same work becomes free raw material for models that answer questions without sending users to the original source.

Why does the legal fight matter beyond one newspaper?

The Times’ lawsuit against OpenAI and Microsoft is widely seen as a test case for the entire publishing industry. If courts decide that training on copyrighted journalism requires permission or compensation, the outcome could reshape the economics of AI development. If not, publishers may have to search for new ways to monetize their archives and reporting.

Sulzberger sees the dispute as bigger than a licensing disagreement. He frames it as a question of whether the AI industry can continue scaling on material created by others while leaving those creators financially damaged.

That issue is especially urgent for journalism, where newsroom budgets have already been under pressure for years. If AI tools reduce traffic to original reporting or absorb its value into chat interfaces and summaries, publishers may find it harder to support expensive reporting operations.

How strong is The New York Times business right now?

By traditional newspaper standards, the Times is in a relatively strong position. Sulzberger said the company has surpassed 13 million subscribers, a milestone that underscores how successfully it has shifted toward a digital subscription model.

He contrasted that with the cost structure of the paper’s journalism, especially overseas coverage and long-term investigations. The business may be healthy, but the reporting itself remains expensive to produce and difficult to justify in narrow financial terms.

That tension is central to the Times’ identity: a strong consumer business is helping support a newsroom that still spends heavily on global coverage, data-driven investigations and original reporting from conflict zones.

What does this say about the future of journalism?

Sulzberger’s answer is blunt: journalism remains essential, but the institutions that produce it are being squeezed from multiple directions. Governments can intimidate, technology companies can disintermediate and the economics of the open web can erode the value of original work.

Still, he argued that the public often underestimates the scale of what major newsrooms do. He noted that the Times has reporters in more than 150 countries and heavy coverage in conflict areas such as Ukraine, Israel-Gaza, Russia, Iran, Lebanon, Sudan, Myanmar and Venezuela. He said the financial case for that level of coverage would be hard to make on a spreadsheet alone.

The argument, then, is civic rather than commercial: societies need reliable reporting even when it is expensive, and especially when it is inconvenient to those in power.

Why does public understanding matter so much?

Because, Sulzberger said, news organizations have not always explained the value of their work well enough. He believes audiences need to better understand what it takes to produce reporting that is independent, verified and often risky.

In his view, journalism does more than describe events. It provides the raw material people use to understand politics, public safety, conflict, business and culture. And because AI systems increasingly ingest that material, he said, the stakes extend beyond media economics into how knowledge itself is assembled online.

That is why he sees the AI debate as inseparable from the future of journalism. If AI depends on reporting to know what it knows, then the rules governing journalism’s use, payment and attribution will affect the quality of the information ecosystem overall.

What happens next in the Times’ AI and legal battles?

There are at least two major fronts to watch. The first is the Times’ copyright case against OpenAI and Microsoft, which is still progressing through motions and discovery disputes. The second is the ongoing clash with the Trump administration over reporter subpoenas and other press-freedom issues.

Each could set an important precedent. The copyright case could influence how AI companies source and compensate content. The press-freedom fight could influence how aggressively future administrations test the boundaries of journalistic protection.

What makes the current moment unusually consequential is that both fights are unfolding at once. The newspaper is trying to defend its reporting, defend the legal rights that protect that reporting and simultaneously decide how to use AI without undermining the work it exists to support.

For Sulzberger, those are not separate problems. They are all versions of the same question: how does journalism survive when powerful institutions, public and private, want its output but not always its independence?

Timeline of the key developments

Bottom line

Sulzberger’s message is that journalism is under pressure from both the state and the machine. The Trump administration, he said, is pushing too far in its efforts to probe and intimidate reporters, while AI companies are building trillion-dollar businesses on the work of publishers and creators without proper compensation.

At the same time, he argues that news organizations cannot afford to respond with fear or passivity. They must defend their rights, keep publishing, and use AI cautiously as a reporting aid rather than a replacement for human judgment.

For The New York Times, that means fighting legal battles, protecting journalists and experimenting with technology all at once — while insisting that the value of independent reporting is worth the cost.

Frequently asked questions

Why is The New York Times suing OpenAI and Microsoft?

The New York Times is suing OpenAI and Microsoft because it says the companies used Times journalism without permission to train AI systems. The paper argues that this practice infringes copyright and threatens the economics of original reporting.

What did A.G. Sulzberger say about AI in journalism?

A.G. Sulzberger said AI should be used as a reporting tool, especially for searching data and spotting patterns, but not as an autonomous writer or publisher. He emphasized that newsroom staff remain responsible for every story and must catch errors themselves.

How many subscribers does The New York Times have?

The New York Times has more than 13 million subscribers, according to Sulzberger. He cited that figure as evidence that the company’s business remains strong even as it faces legal, political and technological challenges.

Why are the subpoenas against Times journalists significant?

The subpoenas are significant because they raise concerns about intimidation of the press. Sulzberger said the Justice Department sought records tied to journalists and even some family communications, which he described as an overreach and a threat to press freedom.

How is AI changing the economics of journalism?

AI is changing journalism economics by reducing reliance on original reporting in some user journeys while also drawing heavily on that reporting as training material. That can weaken traffic, licensing value and audience relationships unless publishers secure compensation or new business models.