CNN sues Perplexity AI as US fragmentation of AI regulation reaches a critical point
Perplexity's communications chief responded to the CNN lawsuit by stating that facts cannot be copyrighted
A single Thursday produced three landmark developments in the world of artificial intelligence law and governance.
Renowned media outlet CNN filed a copyright lawsuit against Perplexity AI, OpenAI published a formal internal governance framework aligned to EU and California law, and the US Department of Justice filed its first-ever federal challenge to a state AI statute in Colorado.
Taken together, the three events represented precisely the outcome the industry had sought to avoid — a fragmentation of artificial intelligence regulation across courts, businesses, and governments, with no unified solution gaining meaningful traction.
Why are publishers taking legal action against Perplexity?
CNN's 54-page filing against Perplexity AI alleges that the company scraped and distributed over 17,000 pieces of its articles, photographs, and videos, using this content as input into AI-generated answers that CNN claims directly compete with its own news output.
The media organisation further alleges that Perplexity falsely implied a content partnership by advertising CNN premium access to subscribers of its Comet Plus tier, despite no licensing agreement existing between the two companies.
Negotiations between CNN and Perplexity had taken place prior to the lawsuit being filed. Perplexity's Chief Communications Officer, Jesse Dwyer, offered the company's position in response, stating that "You can't copyright facts."
CNN, however, maintains that its claim rests on the infringement of copyright in works containing protected expression — specifically its articles, photographs, and videos.
Industry divided on licensing versus litigation
Not all news organisations have chosen the legal route. Outlets including Time, Gannett, Le Monde, and Der Spiegel opted to licence Perplexity rather than pursue litigation.
However, the cost of legally licensed copyrighted training data continues to rise with each new case brought before the courts.
In a related development, the class action suit known as Bartz v. Anthropic — in which plaintiffs alleged that their copyrighted books were used to train AI models without consent — is awaiting court approval of a settlement worth billions in the Northern District of California, expected in mid-May 2026.