https://www.cnet.com/tech/services-and-software/perplexity-ai-cnn-copyright-suit/
The AI search company Perplexity is being sued by CNN and other media companies for copyright infringement.
AI products regularly scrape news publications and websites to answer user questions with real-time data, accelerating the collapse in traffic and revenue to original sources.
In response to the lawsuit, Jesse Dwyer, Perplexity's chief communications officer, told Stetler and other media outlets in a statement: "You can't copyright facts." The US government's Copyright Office states: "Copyright does not protect facts, ideas, systems, or methods of operation, although it may protect the way these things are expressed."
CNN said in its own statement that a company valued at tens of billions of dollars shouldn't "steal from entities that create the original content Perplexity exploits" and that "commercial operators can and must pay to make use of it."
A Perplexity representative didn't immediately respond to a request for comment.
Perplexity is one of several companies, including OpenAI and Anthropic, that have been battling news publishers and media giants over copyright claims.
(Disclosure: Ziff Davis, CNET's parent company, in 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)
More than 100 such lawsuits have been filed. But different conclusions have been reached as to whether training AI models on copyrighted data counts as fair use, said Michael Goodyear, an associate professor at New York Law School. Considerations include how the training occurs, what AI outputs contain and whether there's any competitive harm to copyright holders.
"No appellate courts have yet weighed in on the viability of these copyright infringement claims against AI companies," Goodyear said.
In the CNN case, he said that Perplexity is correct that facts aren't protected by copyright, but the way CNN presents facts could be.
"Even short news articles would typically qualify for copyright protection under the low bar of required originality," Goodyear said. "The question becomes whether the thousands of cases of infringement CNN describes are copying whole paragraphs verbatim, or whether they are paraphrasing or merely copying unprotectable facts."
As plunging website traffic has drained billions in publisher revenue and triggered widespread media layoffs, AI firms are aggravating the crisis. According to a new report from the think tank Open Markets Institute, over the past six months, the rate of AI crawlers bypassing paywalls and blocks has nearly quadrupled, spiking from 3.3% to 12.9%.
That's partly why a number of publishers signed AI content licensing deals with tech companies to monetize content used to train AI systems. One way out for Perplexity may be to renegotiate a licensing deal with CNN. Even if Perplexity has valid legal arguments, a licensing agreement could shift from unauthorized scraping toward a formalized content partnership.
However, the Open Markets Institute report says that when it comes to AI content licensing, news and content creators are trapped in a double bind. The same tech giants whose AI tools are starving websites of human traffic are now the ones gatekeeping the licensing deals meant to replace that lost ad revenue.
(Score: 2, Insightful) by VLM on Wednesday June 03, @01:21PM
CNN is also not an original source that its 65F outside my house right now.
Its a world WIDE web, if 99% of legacy media outlets on the internet collapse we'll still have more choice than we had before the internet, also they're a politically unified, strongly biased, centrally controlled propaganda outlet, so we won't lose any alternative perspectives if someone asks an AI to look up what CNN reported when they read the observations from weather.gov. I will be able to read the observations from weather.gov even if both CNN and the AI company go out of business. I don't need either of them to read it for me then wrap it in ads and pay a subscription.
I journalists give up on journalism and just copy paste press releases, ask AI to write articles, and only ship one parties propaganda via central control, don't be surprised if the public gives up on journalists. What have they ever done for me? Oh nothing, well byeeee maybe they can learn2code like the advice they gave everyone else.
(Score: 5, Insightful) by DadaDoofy on Wednesday June 03, @03:26PM (5 children)
Yes you can freely glean facts from copyrighted material. What you can't do is steal a copyrighted characterization, explanation, summation of, or opinion about the facts, which it seems they've done.
(Score: 5, Interesting) by JoeMerchant on Wednesday June 03, @04:05PM (3 children)
> steal a copyrighted characterization, explanation, summation of, or opinion about the facts, which it seems they've done.
This will be the argument in court, and the AI industry seems to have enough cash to afford lawyers, judges, legislators, PR campaigners...
This will also be a continuing area of development for the AI industry: how to give the user what they are looking for without crossing boundaries of legal definitions of ownership...
I have been rather disappointed in the quality of Google image searches recently. A few years back, you could ask for images of XYZ and then sub-set the results selection by copyright status - seeing only free-to-copy images if that's what you wanted. That functionality has apparently been nerfed lately.
🌻🌻🌻🌻 [google.com]
(Score: 5, Insightful) by aafcac on Wednesday June 03, @04:39PM (2 children)
I've got mixed feelings here. I don't really want an expansion of copyright rights, but by the same token if AI companies are allowed to take information that cost money to obtain and give it away, it's just going to accelerate the rate at which the media degenerates into paid advertising and fake news. It wasn't that long ago that CNN was largely reliable as a news source, now you've got people on both the left and the right calling them out for making stuff up and failing to really do much to spread accurate information on key issues.
(Score: 1, Informative) by Anonymous Coward on Wednesday June 03, @04:46PM
This is as designed.
(Score: 4, Touché) by JoeMerchant on Wednesday June 03, @07:23PM
>accelerate the rate at which the media degenerates into paid advertising and fake news.
I feel like there must be some asymptote that we started approaching a few years back, well over 50% heavily spun or outright false seems to have been the norm for over a decade now.
🌻🌻🌻🌻 [google.com]
(Score: 3, Insightful) by looorg on Wednesday June 03, @04:55PM
If you can't copyright fact then you really can't copyright anything. The product, whatever it might be, is the fact or result of whatever procedure have been done. This is just AI companies and AI executives being or coming out as the pond sucking slime that they are. Huckster the lot of them.
(Score: 1, Interesting) by Anonymous Coward on Thursday June 04, @03:00AM
How about stuff like Faux News? If they make up a story, and you copy it and rewrite it in your own words how much rewriting do you need till it no longer counts as infringement?
Compare:
https://en.wikipedia.org/wiki/Rogers_v._Koons [wikipedia.org]
https://12tomatoes.com/kimba-similarity-lion-king/ [12tomatoes.com]
https://thefilmera.com/2018/09/04/the-magnificent-seven/ [thefilmera.com]