A.G. Sulzberger is not apologizing for spending $20 million to fight OpenAI and Microsoft. To most publishers, that kind of cash vanishing into legal limbo would trigger panic. For the New York Times publisher? It’s just the price of doing business.
“It’s the cost of defending journalism,” he told me in late July. “At the Times and everywhere else.”
Sulzberger has held the helm for roughly a decade now. He’s steered the paper through the digital revolution. But lately? He’s defending the very concept of news itself.
The lawsuit against AI giants is only half the war. The Trump administration is actively trying to subpoena reporters. They served subpoenas directly to the homes of Times journalists who broke the story on Qatar-gifted Air Force One. They subpoenaed third-party phone providers. They went after family members.
It is an escalation unlike anything seen in over a century.
“This is the biggest effort to go after thepress that we’ve seen from a White House… since Woodrow Wilson.”
The timing is brutal. The Times recently hit 13 million subscribers. It is flush with cash compared to the rest of the industry. Its newsroom is stable. Yet Sulzberger sees existential dread everywhere.
AI companies are eating the open web. The government is attacking reporters. And other media companies are collapsing under the weight of their own digital disruption.
We need to talk about how the New York Times is handling AI in reporting. Not as a sci-fi concept. But as a tool in the newsroom today.
How The New York Times Uses AI in Reporting
Most outlets are messy about this. Some handed full article drafts over to bots. That was foolish. Mistakes were published. Fake journalists were invented.
Sulzberger calls it “a completely foreseeable and unnecessary set of errors.”
The Times approach is different. They treat AI as a filter, not a writer.
Consider a reporter who recently won a Pulitzer for AI-supported reporting. He was analyzing bomb craters in Gaza and Israel. He had millions of satellite images. AI helped narrow that down to thousands.
Then the human does the work. He looks at every single one of those thousands. He checks for anomalies. He confirms the pattern.
“We are using [AI] much more in reporting than in writing, crafting, or editing,” Sulzberger says.
Think about campaign finance. That is essentially pattern recognition. Who are the donors? What are the shifts in big donors versus small ones? Where are the anomalies? AI excels at spotting that noise.
“You have to think of AI as a ‘filter.'”
It shrinks the dataset. It highlights the signal. But the journalist remains responsible. Always.
“You would never ask a computer, ‘What if it makes a mistake?’ You’d fix the bug,” he argues. “If AI makes a mistake, it’s your job to find it and take it out.”
The philosophy here is patience. Most news orgs jumped in too fast. They wanted to be first. The Times decided to wait, watch, and study. They experimented quietly behind the scenes.
Why? Because being right beats being first.
The Chilling Effect on Journalism
Sulzberger doesn’t mince words about the political climate. He calls the Justice Department’s actions “gross overreach.”
This isn’t just about the Times. Colleagues in Washington are reporting something worse.
They say other newsrooms are self-censoring.
Some reporters are telling their editors they won’t pursue certain stories. Not because they lack interest. But because they are terrified of retaliation. Their strategy? Wait for the Times or The Wall Street Journal to run the story. Then follow it up.
They see the Times as the shield. Breaking the story is too risky.
“That is the definition of chilling,” Sulzberger says. “The message that retaliation must be avoided gets down to the lowest level of the newsroom.”
His job as publisher? Break that cycle.
He called each subpoenaed reporter personally. He told them: “We have your back. We’ll do whatever we need to support you.”
Their response?
“Oh, we’re not going to let go of this story.”
Good.
Because holding your ground doesn’t mean being adversarial. It means publishing the truth when powerful people get angry. It means asking the hard questions.
The industry has largely held its line. The Associated Press sued over the “Gulf of America” naming dispute. They didn’t sit down. They fought.
Your rights only hold if you enforce them. Avert your eyes, drop your head? You lose everything.
Why Independent Journalism Matters More Than Ever
Sulzberger admits the media has failed at one thing: explaining why this work matters to the public.
We’ve taken it for granted.
The Times puts reporters in 150 countries every year. Last year that included Myanmar. Sudan. Lebanon. Iran. Venezuela. Gaza. Russia.
Ukraine alone hosted 70 Times journalists on the ground.
Is there a financial case for that? Maybe not. If you crunch the numbers on months of investigative work yielding a single story? The ROI looks thin.
But the civic value is incalculable.
Thomas Jefferson put it best:
“Given the choice between a government without a press, or a press without a government, I would not hesitate to choose the latter.”
Journalism is how you know what you know.
And now? It’s how AI knows what it knows.
AI models are trained on our reporting. They consume our facts. If we stop producing high-quality journalism, the machines get dumb. The information ecosystem collapses.
The Times is betting $20 million on that principle.
Sulzberger isn’t asking for pity. He’s asking for clarity.
We are at a pivot point. The government is targeting the press. AI is reshaping consumption. And the public is still scrolling past the headlines.
Will the Times save journalism? Maybe not on its own.
But by refusing to blink? By forcing AI companies to the negotiating table? By protecting the reporters getting served subpoenas at their front doors?
They are setting the precedent.
The question isn’t whether the Times can save the industry. It’s whether the rest of the industry will learn how to stand with them.
Or if they’ll keep waiting for the Times to take the bullet first.
We’ll see. The reporters won’t stop. But the public? They need to understand that when the work stops, we’re all flying blind.
And no amount of AI can fix that.
