AI search tools are changing the economics of online publishing by giving users direct answers instead of sending them to the original source. For readers, that shift can feel seamless and efficient, but publishers have concerns.
CNN recently sued Perplexity in federal court, alleging that the AI search company unlawfully copied and distributed CNN stories and related works through its AI-powered answers. CNN’s position is that Perplexity is not merely helping users locate information – it is allegedly using CNN’s protected journalism to power and train a competing platform without authorization or payment.
Perplexity’s response points to one of the oldest principles in copyright law: facts are not copyrightable. That principle is true and important. For example, no one owns the fact that an election happened. Public facts must remain available for ordinary public knowledge.
But journalism is rarely just an assortment of facts. It requires editorial judgment at every stage and a reporter’s work and effort to research and present the story. Copyright does not protect the underlying facts themselves, but it can protect the journalistic work.
The Issue Is Not Whether AI Can Know Facts
The public conversation around AI and copyright often is reduced to the idea that facts are free. While that statement is legally correct, it does not resolve the harder question presented by AI answer engines like Chat GPT or Claude AI.
The legal issue begins when an AI tool allegedly copies protected expression rather than merely using factual information. CNN’s lawsuit argues that Perplexity copied and reused thousands of CNN works to support its products, including original written stories, videos and images. CNN also alleges that Perplexity’s answers can act as a substitute for the original material, keeping users inside Perplexity’s platform instead of sending them to the source, CNN.
Why “Facts Are Not Copyrightable” Is Only the Beginning
Perplexity’s reported position that facts cannot be copyrighted is legally sound, but it is not the end of the analysis. Copyright law has long recognized a difference between facts and expression. The Supreme Court made this clear in Feist Publications v. Rural Telephone Service Co., a case involving telephone directory listings. 499 U.S. 340 (1991). There, the Court held that facts themselves are not protected by copyright, but a sufficiently original selection or arrangement of facts may be protected.
That principle is especially important in journalism because a news article may contain public facts while still reflecting original expression. Copyright does not protect the raw facts, but it can protect the storytelling that presents those facts.
This is where AI answer products create tension. If an AI tool provides a short factual answer based on general knowledge, the copyright issue may be limited, but if it reproduces protected language, closely paraphrases original reporting, summarizes paywalled work, or displays images and video without authorization, the legal risk increases.
The Problem With Replacement
The CNN case is not only about copying. It is also about substitution, which may become one of the most important issues in AI copyright litigation. Traditional search engines like Google work by displaying snippets and display links that send the user to the source website. However, AI answer engines work differently by showing the user a complete answer within their own platform. The user receives the information, but may not ever click through to the original source.
For publishers like CNN, AI answer engines create a serious challenge because they can reduce the need for users to visit the original news site. That matters because website traffic is one of the key metrics publishers rely on to attract advertisers and generate revenue.
The larger issue is that reporting and journalism are expensive. Newsrooms require reporters, editors, fact-checkers, legal review, travel, research, and time. If publishers lose too much traffic and revenue, they may be forced to reduce the reporting teams that produce the original journalism AI systems rely on.
In other words, AI depends on quality information, but it may also be disrupting the business model that makes quality information possible. This may not be an immediate crisis today, but over time it could affect how news is reported, how facts are verified, and how the public accesses reliable information.
The Expanding AI Content Problem
The lawsuit between CNN and Perplexity also points to a broader issue: AI copyright disputes are not limited to written words. CNN’s allegations include stories, videos, images, and related works, which reflects the way modern media companies actually operate. Journalism today is responsible for an array of multimedia across different platforms, from articles to live coverage and digital packages.
Answer engines can be used to create content at scale, and often, that content is reused but may not have permissions. The use of answer engines within companies can increase efficiency, but it can also make copyright mistakes easier to scale.
The Business Takeaway From CNN v. Perplexity
If your company publishes original content online, that content has value. Blog posts, white papers, product photography, videos, training materials, reports, and guides may all carry copyright interests. The fact that content is publicly accessible does not mean it is free for anyone to copy or repurpose commercially.
At the same time, businesses using AI tools must exercise caution. AI can help with research, drafting, brainstorming, and content organization, but it should not be construed as permission to reuse protected material. If an AI tool produces language that closely tracks a source, lifts images, or summarizes paywalled content in a way that replaces the original, that can create risk.
At Minx Law, we see this as one of the most important practical issues for modern businesses. AI is not going away, and neither is copyright law. Companies need policies that account for both.
That may mean setting internal rules for AI-assisted writing, training teams not to paste third-party articles into AI tools without permission, reviewing AI-generated content before publication, and protecting original company content through registrations, licensing terms, and enforcement strategies. The goal is not to avoid AI altogether; rather, the goal is to use it with judgment.
