CNN thinks facts belong to them. Perplexity thinks facts are free for the taking.
This clash isn’t about abstract philosophy. It’s a federal lawsuit filed Thursday in a New York federal court. CNN claims the San Francisco search engine Perplexity stole their livelihood. Specifically, the network accuses the startup of copying more than 17,000 CNN stories. Videos too. Images. Whole packages of original reporting distributed without asking, let alone paying.
Jesse Dwyer, Perplex’s chief communications officer, issued a blunt response.
You can’t copyright facts.
He’s right. Sort of. The US Copyright Office agrees that raw data isn’t protected. But how that data is dressed up? That’s a different story.
Why CNN is Suing an AI Startup Over Copyright
CNN isn’t playing around. This is their first legal shot across the bow of the AI industry. But they are walking into a crowded courtroom. The New York Times has sued. News Corp has sued. The pattern is clear.
Publishers want a seat at the table. They want payment for the ingredients they provided. CNN tried to negotiate a licensing deal with Perplexity first. It failed. Talks stalled. No agreement was reached.
This isn’t a vacuum. Last year, CNN struck a deal with Meta. Tech giants like Facebook’s parent company know how to compensate media outlets. Meta pays CNN to use its content for Meta AI answers. Why doesn’t Perplexity pay too? That is the core of the dispute.
- CNN claims unauthorized copying of verbatim content.
- Perplexity relies on fair use and public data.
- Publishers see traffic collapsing under AI crawler weight.
The stakes are financial and existential. If startups can scrape entire newsrooms without paying, the traditional business model crumbles. If courts side with the startups, publishers must pivot to selling access directly to bots, not just humans.
Can You Copyright Facts and AI Summaries?
This is where it gets sticky. Michael Goodyear, who teaches at New York Law School, points out a critical nuance. Yes, facts themselves are not copyrightable. True.
But CNN articles aren’t just lists of facts. They are original expressions.
“Even short news articles would typically qualify,” Goodyear explains.
The legal threshold is low for originality. The question is not whether Perplexity used CNN’s facts. The question is whether it copied CNN’s expression of those facts. Did it lift paragraphs whole cloth? Or did it merely summarize?
Courts are still figuring this out. No appellate courts have finalized a verdict on whether training AI on copyrighted text constitutes fair use. The jury, metaphorically, is out.
The High Cost of AI Licensing Deals for Media
The underlying problem is simpler. AI crawlers are eating publisher traffic.
A recent report from the Open Markets Institute highlights the severity. Over the last six months, the rate of bots bypassing paywalls quadrupled. We jumped from 3.3 percent to nearly 13 percent. These crawlers replace human visitors who would otherwise click ads. Revenue vanishes.
Publishers face a trap.
1. Block bots, but risk missing out on potential licensing fees.
2. Let them scrape, but lose ad income.
3. Negotiate licensing deals, but beg at the gate of the companies causing the problem.
This is a double bind. The tech giants destroying ad revenue are the same ones holding the purse strings for licensing agreements. It creates a dependency that feels like extortion.
One path out exists: renegotiation. If Perplexity faces overwhelming legal liability, a licensing deal might look smarter than litigation. It turns an enemy into a partner. But for that to work, publishers need leverage. They need proof that the damage is significant and that they control something valuable.
Is CNN right to sue? Probably not for every single sentence. But maybe they are right that the system is broken. We have yet to decide who pays for the information age.
