Publishers Have Split Into Two Camps on AI, and News Corp Is in Both
Three years into the argument about AI and content, publishers have sorted themselves into two strategies. One camp licenses. The other sues. A few run both at once, and those are the ones worth studying.
News Corp is the clearest example. It licensed to OpenAI in a deal reported at up to 250 million dollars over five years, signed a further arrangement with Meta reported at up to fifty million a year for three years, and simultaneously sued Perplexity through Dow Jones and the New York Post, describing what it alleged as content kleptocracy. It has since taken action against Brave as well.
That’s not inconsistency. It’s segmentation.
The logic of doing both
An AI company that will pay is a customer. An AI company that won’t is an infringer. The same content, the same use, and the difference is whether there’s a contract.
Which sounds cynical until you notice it’s how every rights business has always worked. Music publishers license to the platforms that pay and litigate against the ones that don’t. Sports leagues sell broadcast rights and send takedowns to everyone streaming without one. Nobody calls that hypocrisy because the framework is established.
The AI market has no established framework, so the same behaviour reads as opportunism. It isn’t. It’s a rights holder discovering that its leverage varies by counterparty and acting accordingly.
What the deals actually look like
The public numbers are uneven and mostly undisclosed. OpenAI has signed with close to twenty media groups, including the Washington Post earlier this year. Perplexity started with a revenue-sharing publisher programme, backed by a pool reported around 42.5 million dollars, paying participating publishers when their content is used in an answer, and has since added conventional licensing agreements too.
Two models, and they treat publishers very differently. A flat licence pays a known sum regardless of usage, which suits a large publisher with negotiating power and steady output. Revenue sharing pays according to citation, which suits nobody yet, because the pools are small, the attribution is opaque and the per-citation economics have never been disclosed in a form anyone can audit.
Ask a participating publisher what a citation is worth and watch how the answer arrives.
The camp that can’t choose
Most publishers aren’t in either camp. They’re too small to be worth a licensing negotiation and too small to fund multi-year litigation against a company with unlimited legal budget.
That’s the majority of the industry. They get scraped, they get summarised, they lose the traffic, and their options are to join a collective action, block the crawlers and lose discoverability, or accept it.
Collective licensing is the only structural answer, and it keeps stalling. Publishers compete with each other, jurisdictions differ, and any pool large enough to matter starts to look like coordinated pricing to a competition regulator. The organisations attempting it are making progress slowly, in an environment where slow is the same as never.
Litigation is producing deals, not doctrine
The pattern to watch is settlement. A Brazilian publisher’s suit against OpenAI ended in a commercial agreement. That’s the template: sue, negotiate, convert the claim into a contract, drop the case.
Each conversion removes a case that might have produced a ruling. What accumulates instead is a set of private prices, known only to the parties, that establish nothing for anyone else.
The industry keeps waiting for a court to define what content is worth. The market is defining it first, one confidential settlement at a time, and small publishers won’t see the terms.