Big Media’s AI Deals—And What Comes Next
The New York Times made headlines earlier this month by cutting a deal with Amazon, letting the tech giant use its articles to train AI models. Financial terms weren’t disclosed, but it’s a sharp pivot for the paper—just months ago, they sued OpenAI and Microsoft for doing the same thing without permission.
It’s part of a trend. Media companies, wary of lawsuits but also eyeing new revenue, are signing licensing agreements left and right. OpenAI’s done deals with Axel Springer (which owns Politico and Business Insider), the Financial Times, and even Vogue’s parent company. For publishers, it’s a way to monetize archives. For AI firms, it’s a shield against copyright claims—and access to reliable data.
But here’s the thing: these deals are happening behind closed doors. Smaller publishers or independent creators? They’re often left out.
Could Decentralization Fix This?
Some argue there’s a better way. “Closed models win short-term sprints,” says Phil Mataras of AR.IO. “Decentralized models win the marathon.” The idea is simple: instead of private, one-off deals, content could live on open networks where usage is tracked transparently. Smart contracts could handle payments automatically—no middlemen, no secrecy.
Aaron Basi from IoTeX Network points out that blockchain-based systems could let creators set their own terms. “It’s still early,” he admits, but for smaller outlets, it might level the playing field. Tokenization, for example, could help track how content’s being used—and ensure creators get paid.
There are already tools for this. Blockchain provenance systems can log a dataset’s entire history—who created it, who bought it, when. Watermarking tech makes theft harder. And decentralized autonomous organizations (DAOs) could let groups of creators negotiate collectively, like a digital union.
The Trade-Offs
Licensing deals aren’t going away. They’re straightforward, and for big publishers, they guarantee income. But they’re also opaque. If you’re a reader, you might never know if an AI’s response was trained on paywalled news. If you’re a small blogger, good luck getting Amazon to return your calls.
Basi thinks transparency will win in the long run. “People want to understand what goes into the tools they use,” he says. Open-source models, while slower to start, might gain trust precisely because they’re not hiding anything.
For now, though, the money’s in closed-door deals. The question is whether that’s sustainable—or if the next wave of AI will demand something more open.
