The New York Times isn't just suing over stolen articles anymore. Now they're saying Microsoft built the whole damn machine — a supercomputer designed to rip off copyrighted work at industrial scale.
Last week, the paper filed an amended complaint in its copyright lawsuit against OpenAI and Microsoft, and let me tell you, this thing has teeth. The new filing claims Microsoft constructed a supercomputer specifically to train OpenAI's models on copyrighted material — and knew exactly what it was doing.
This isn't just a legal tweak. It's a strategic pivot, timed perfectly after the Supreme Court handed down a ruling that shook the copyright world.
What the Supreme Court Changed
Earlier this month, SCOTUS ruled against Sony in a case that limited fair use protections for tech companies that copy massive amounts of data. The decision sent shockwaves through the AI industry, which has long leaned on fair use as its get-out-of-jail-free card.
Now the NYT is using that ruling like a crowbar. The amended complaint argues that Microsoft's supercomputer — built with tens of thousands of GPUs and custom networking — wasn't just a neutral tool. It was a deliberate infrastructure designed to ingest copyrighted news articles without permission.
“Microsoft didn't just write a check,” the complaint states. “They built the engine that powers this infringement.”
The Supercomputer Nobody's Talking About
Details about the machine are sparse, but sources familiar with the project say it's one of the largest supercomputers on the planet. Microsoft invested billions to create a cluster capable of training models like GPT-4. The system is optimized for massive data throughput — exactly the kind of hardware you'd need to scrape and process the entire internet, including paywalled content.
The NYT claims Microsoft knew the data being fed into that machine included copyrighted Times articles. Emails and internal documents reportedly show engineers discussing how to avoid detection or legal blowback.
“They built a copyright-infringement factory,” said a legal expert who reviewed the filing. “If the court buys this argument, Microsoft is in deep trouble.”
OpenAI Plays Dumb, Microsoft Plays Silent
OpenAI has consistently argued that training on publicly available data falls under fair use. But the NYT's new approach shifts the focus to Microsoft, which has deeper pockets and a more explicit role in the infrastructure.
Microsoft declined to comment on the specific allegations, but a spokesperson reiterated that the company respects intellectual property rights and believes its AI partnerships are lawful. OpenAI didn't respond to requests for comment.
Meanwhile, the NYT isn't backing down. The paper is asking for damages that could run into the billions, and perhaps more importantly, an injunction that would force Microsoft to dismantle or repurpose the supercomputer.
“If the NYT wins, every tech company building AI infrastructure will have to rethink how they handle training data. This could reshape the industry.”
Why This Matters Beyond the Newsroom
This isn't just a fight between a newspaper and a tech giant. It's a test case for the entire generative AI economy. If Microsoft can be held liable for building supercomputers that train on copyrighted data, then Google, Amazon, and every startup with a GPU cluster faces the same risk.
The entertainment industry is watching closely. Hollywood studios, music labels, and book publishers have all sued AI companies over training data. But none have gone after the hardware provider directly.
“This is brilliant lawyering,” said a copyright attorney not involved in the case. “By targeting the infrastructure, they're going after the one thing AI companies can't easily replace.”
The Human Cost
Lost in the legal jargon is the fact that real journalists wrote those articles. Real reporters spent weeks investigating stories, only to have their work scraped into a machine that spits out cheap imitations.
The NYT employs over 1,700 journalists. Many of them watched their bylines get fed into an algorithm without consent or compensation. This lawsuit isn't just about money — it's about whether human creativity can survive in a world where machines can replicate it for free.
One veteran reporter told me: “We're not Luddites. We use AI tools in the newsroom. But if they can just take our work and sell it back to us as a chatbot, what's the point of paying us?”
What Happens Next
The case is in the early stages. Discovery will likely reveal more emails, more technical specs, and more embarrassing admissions. Microsoft will fight hard — this case threatens not just the OpenAI partnership but its entire Azure AI business.
The NYT has already won round one by shifting the narrative. Now it's about proving that a supercomputer can be a tool of copyright infringement, not just a neutral box of circuits.
If they succeed, the AI industry will never be the same. And maybe that's exactly what needs to happen.
You want to know who built the machine that's eating the internet? Microsoft did. And they're about to pay for it.
