Chamath Palihapitiya, former Facebook executive, venture investor and co-host of the All-In podcast, says copyright may soon be unenforceable in an artificial intelligence-driven world and urges companies to build competitive advantages that do not depend on protections that could quickly lose all value. As for artists, he questions whether they need copyright to earn a living, suggesting they can earn income through other means, such as live performances. This is a convenient stance for someone running or investing in businesses that require a massive amount of data for training and develop advanced AI models.
The AI Copyright Divide
U.S. copyright law holds that only people can be authors. The Copyright Office and courts have said works created entirely by AI don’t qualify because they lack human creativity, even if a person wrote the prompt. But when humans use AI tools and add meaningful creative choices, their work may still be protected.
Fair use is decided on a case-by-case basis. Judges look at why the material was used, how much was taken, and whether it affects the original’s market. AI training on copyrighted works may pass if it transforms the content and avoids undercutting sales. In other situations, courts could find it infringes. There is no blanket rule, and disputes are likely to be settled in court for years.
Palihapitiya’s perspective, while provocative, reflects a growing skepticism among some technologists about the durability of traditional intellectual property laws in the face of generative AI. The central thesis is that the power of AI to independently create or derive from existing works will render existing legal protections obsolete. Why would a copyright hold up when an AI could theoretically generate a functionally identical or even superior work without ever having copied the original? This line of thinking suggests that the traditional legal framework, built on the premise of human-to-human creative exchange, is ill-equipped to handle the age of machine learning. The idea is that AI models learn patterns and create new material rather than copy existing work. This distinction may be the undoing of copyright as we know it.
The counterargument, however, is made by Jason Calacanis, another host of the podcast, and a defender of copyright holders. Calacanis argues that the rights of creators, such as journalists, musicians and filmmakers, must be protected. He believes that AI companies, which are becoming some of the most valuable corporate entities in history, should not be allowed to build their fortunes on the work of others without proper compensation. He envisions a future where licensing agreements become a foundation of the new AI economy. In this scenario, copyright does not die; instead, it evolves into a revenue stream, creating a golden era for content creation by funding more journalists, artists and fact-checkers.
Training AI Data: Fair Use Of Copyrighted Material?
The debate also centers on the legal and conceptual distinction between training data and output. David Friedberg and David Sacks, the other two podcast hosts, offer a more balanced perspective to bridge the divide. They both favor a fair use interpretation for AI training. Friedberg believes that if information is available on the open internet, an AI learning from it is no different from reading books to learn. Sacks agrees, drawing a clear line between the training process and the final output. They argue that the infringement occurs only if the AI’s output is a direct copy or plagiarism of copyrighted material. This approach attempts to strike a balance, allowing AI to improve by learning from a broad pool of sources while still holding the AI and its developers accountable for plagiaristic outputs. This was the perspective defended by President Trump during the announcement of the AI Action Plan, highlighting Sacks’ influence as the chair of the President’s Council of Advisors on Science and Technology.
A Case Study In AI’s Copyright Clash
A similar debate recently played out in a public dispute between internet infrastructure giant Cloudflare and AI search engine Perplexity. The conflict encapsulates the competing viewpoints. Cloudflare accused Perplexity of using stealth crawling to access websites that had blocked them. Web crawling is a practice where a computer program automatically browses websites to collect information, much like a super-fast, automated visitor. Stealth crawling is when a web crawler hides its identity to bypass restrictions and reach sites that have barred it. Cloudflare’s CEO, Matthew Prince, warned that AI models pose an existential threat to publishers’ business models, echoing Calacanis’s view that creators’ livelihoods are at risk.
Perplexity, however, pushed back, denying the accusations and characterizing the incident as a misunderstanding of its technology. The company argued that its AI assistants operate on behalf of a user’s request and are not systematically crawling the web for training data. This defense reflects Sacks and Friedberg’s view that the key issue is the AI’s purpose and how it’s applied. Perplexity is fulfilling a specific, user-driven function, not stealing to build a new model, which they claim is a fundamentally different action.
The dispute highlights the point that traditional rules are becoming fragile. Cloudflare claims its efforts to block certain automated programs, a bedrock of internet etiquette, were circumvented. This situation shows the difficulty of enforcing old rules when new technology can so easily bypass them. It also highlights the different philosophies: is an AI bot an uninvited guest or a helpful assistant working on a user’s behalf?
This is a clash between two visions of the future. One embraces a world where technology moves faster than the law, forcing us to abandon old notions of ownership in favor of new, more resilient business models. The other sees a future where copyright is not an outdated legal concept but a vital economic engine that can be adapted and monetized in the age of AI. The middle ground points us to a path forward that adapts current laws to fit AI’s real-world usage.
The future of copyright is unlikely to be a simple binary decision. Instead, it will be a negotiation between creators, tech companies, lawyers, and regulators. What’s clear is that the conversation is no longer confined to legal journals but has entered the mainstream, sparking a necessary dialogue about the value of creativity, the nature of intelligence, and the future of the digital economy.