Copyright Claims Expose the Undervaluation of Creative Labor

The latest revelations in The New York Times’ lawsuit against OpenAI and Microsoft have laid bare a troubling truth: in the rush to build transformative AI systems, the value of human creativity and journalistic labor has been systematically discounted, if not entirely dismissed.

Newly unsealed court documents in the ongoing copyright dispute between The New York Times and artificial intelligence giants OpenAI and Microsoft reveal far more than alleged technical violations. They expose a fundamental tension at the heart of the AI revolution – the clash between technological innovation and the protection of creative labor. Internal communications suggest that key figures at both companies were aware of methods to circumvent paywalls and access copyrighted content, yet proceeded with training their large language models on this material. More significantly, the documents reveal that employees within these organizations recognized the existential threat AI posed to publishers, with one Microsoft executive describing the practice as “an astonishing theft of unprecedented proportions” and potentially “the largest theft of labor in human history.”

This case transcends the legal technicalities of copyright infringement. It serves as a mirror reflecting society’s broader struggle to assign value to creative and intellectual work in an era where technology promises efficiency and scale at the expense of individual creators. The unsealed filings paint a picture of companies that understood the implications of their actions yet moved forward, prioritizing innovation over compensation and collaboration.

At the center of the latest revelations is an alleged exchange between OpenAI researcher Nick Ryder and co-founder Greg Brockman regarding methods to access content behind The New York Times’ paywall. According to court filings, Ryder informed Brockman about a way to bypass the newspaper’s subscription barrier, to which Brockman allegedly responded, “ah nice.” While these allegations have not yet been established as court findings and some exhibits remain sealed, the implication is clear: there was awareness within OpenAI’s leadership of efforts to access protected content.

Paywalls exist for a reason. They represent the economic model that sustains quality journalism in the digital age. When print advertising revenues collapsed with the rise of the internet, news organizations turned to digital subscriptions as a lifeline. The paywall became the barrier protecting the labor of reporters, editors, fact-checkers, and photographers – the humans who invest time, expertise, and often risk to create original reporting. By allegedly circumventing these barriers, AI companies weren’t just violating terms of service; they were undermining the very economic structures that make investigative journalism and quality reporting possible.

The revelation about paywall workarounds is particularly significant in the context of fair use, the legal doctrine that OpenAI and Microsoft have relied upon in their defense. Fair use typically applies to publicly available information and allows for limited use of copyrighted material for purposes such as criticism, commentary, or education. However, when content is deliberately hidden behind a paywall – requiring users to pay for access – it becomes much harder to argue that scraping that content constitutes fair use. As one legal expert quoted in background research noted, paywall violation “eviscerates” the fair use defense.

Microsoft CEO Satya Nadella’s testimony, as cited in the court filings, appears to acknowledge this principle. According to the documents, Nadella stated he would have required OpenAI to retrain its models if he had known that paywalled information had been scraped and incorporated into training data. He also reportedly said that paywalled material should be licensed when used for AI training. While Microsoft has clarified that these remarks should not be interpreted as a legal conclusion on the copyright dispute, they suggest an understanding that content creators deserve compensation when their protected work is used for commercial purposes.

Newly unsealed filings in The New York Times’ copyright case against OpenAI and Microsoft broaden the debate from training data to the economics of creative labor. Allegations about paywall access, large collections of publisher material and sharp reductions in referral traffic suggest that AI systems may not only learn from creative work but also compete with the institutions that finance it. The case remains unresolved, but it raises a central question for the digital economy: can AI scale without treating human reporting, research, art and writing as cost-free inputs?

Perhaps the most striking element of the unsealed documents is the evidence of internal concern within Microsoft itself. Brent Hecht, Microsoft’s director of applied science, is quoted in the filing describing AI training on copyrighted content as “an astonishing theft of unprecedented proportions” and potentially “the largest theft of labor in human history.” These are not the words of an external critic or competitor; they come from within the organization actively developing and deploying these AI systems.

Hecht’s characterization frames the issue in starkly moral terms. By calling it “theft of labor,” he connects the practice to fundamental questions about how society values work. Journalism, writing, photography, and other creative endeavors require skill, time, and often significant investment. When AI systems are trained on millions of copyrighted articles without permission or compensation, the labor that produced those articles is effectively appropriated without recognition or payment.

The filing also indicates that Hecht warned of what he termed an “accidental cover up” as efforts were made to identify material from The New York Times and other plaintiffs inside AI systems. This suggests that once the legal challenges emerged, there may have been attempts to assess the extent of copyrighted material within training datasets – efforts that Hecht apparently feared could obscure the full picture of what had occurred.

Microsoft has been careful to distance itself from Hecht’s comments, stating they reflected his individual perspective rather than the company’s legal position. This distinction is revealing. It acknowledges that serious ethical concerns existed within the organization while simultaneously maintaining that these concerns don’t represent corporate policy or legal liability. The gap between individual conscience and corporate action highlights a recurring pattern in technology companies: employees may recognize problematic practices, but institutional momentum and commercial interests often prevail.

The unsealed documents also provide concrete evidence of what many in the publishing industry have long feared: AI-powered search and chat tools pose an existential threat to news organizations by reducing traffic to their websites. According to internal Microsoft data cited by The New York Times, the company’s Copilot answer-focused search experience could reduce click-throughs to the Times’ website by as much as 93% compared with conventional Bing search results.

This statistic is devastating. For digital publishers, web traffic directly correlates to revenue through both advertising and subscription conversions. When users receive AI-generated summaries of news content without clicking through to the source, publishers lose both the immediate ad revenue from that visit and the potential to convert casual readers into paying subscribers. A 93% reduction in traffic would be catastrophic for any news organization, regardless of size or prestige.

OpenAI executive Nick Turley, who leads ChatGPT, is quoted in the filing describing the impact on publishers as an “existential threat.” The filing indicates Turley warned that AI products were already capable of replacing some visits to publisher websites and could become more substitutive as the technology developed. This acknowledgment is significant because it comes from a senior leader at OpenAI, the company at the center of the generative AI revolution. Turley’s comments suggest that the competitive threat to publishers was not only foreseeable but actively recognized within the organization.

The Times’ lawyers are using this data to challenge the argument that AI products merely transform publicly available information without competing with publishers. If AI systems are trained on copyrighted content and then provide that content in ways that directly substitute for visits to the original publishers, the fair use defense becomes much weaker. Fair use requires that the new use be transformative and not serve as a market substitute for the original work. When an AI chatbot can provide users with the essential information from a news article without them ever visiting the publisher’s site, it’s difficult to argue the use is not substitutive.

The strongest lesson from the filings is that efficiency cannot be measured only by how quickly an AI system answers a user. If the system removes the visits, subscriptions and licensing revenue that support original work, it may be shifting costs rather than eliminating them. AI can assist journalism and other creative fields, but a durable ecosystem needs clear permission, meaningful attribution and compensation that reflects the value of the labor being absorbed.

The court filings also reveal the sheer scale of alleged copyright infringement. According to The New York Times, OpenAI’s mid-training datasets contained more than 91,692 copies of works from The New York Times, Daily News, and the Center for Investigative Reporting. Another dataset derived from Common Crawl allegedly contained more than two million documents originating from nytimes.com alone.

These are not incidental or accidental inclusions. They represent a systematic collection of copyrighted material on a massive scale. The filing also references Project Mango and Project Taxi, initiatives through which Microsoft and OpenAI allegedly exchanged or assembled training material. Project Mango reportedly contained copies of at least 160,903 distinct works produced by publishers involved in the litigation.

Adding another layer to the allegations, the Times claims that researchers removed copyright notices from some training material because they did not want those notices appearing in AI-generated responses. If proven, this would suggest a deliberate effort to obscure the origins of copyrighted content, making it harder to trace AI outputs back to their sources.

The scale and alleged deliberateness of these actions undermine the narrative that AI companies simply used publicly available data in good faith. Instead, the picture that emerges is one of systematic appropriation of copyrighted works, awareness of the legal and ethical problems this created, and continued development despite internal warnings.

Beyond the specific legal claims, this case raises fundamental questions about how society values knowledge and creative labor. For centuries, the production of journalism, literature, music, and art has required either patronage, direct payment, or indirect compensation through advertising and licensing. The digital revolution disrupted many of these models, but subscription and advertising-supported content remained viable paths for creators to earn a living.

AI threatens to break this connection entirely. If AI systems can be trained on copyrighted works without compensation and then generate content that substitutes for the originals, the economic model for creative production collapses. Why would a rational actor pay for a New York Times subscription if an AI chatbot trained on Times articles can answer questions and provide information for free?

This is not a hypothetical concern. Multiple publishers have already joined the lawsuit against OpenAI and Microsoft, with a coalition now exceeding 550 publications. These range from major metropolitan newspapers like The Seattle Times to smaller local outlets. Their collective action reflects a recognition that the stakes extend beyond any single organization. As one filing stated, “Unless Defendants are held accountable for stealing, stripping and misusing the publishers’ content, the AI boom Defendants orchestrated and benefit from will be a death knell for local journalism.”

The implications extend beyond journalism. If AI companies can freely appropriate creative labor for training purposes, writers, artists, musicians, and other creators face similar threats. The precedent set in this case will likely influence how copyright law applies to AI development across all creative industries.

A careful discussion of AI and copyright should avoid these recurring analytical mistakes.

At its heart, this conflict reflects a tension that runs throughout the history of technological change: the balance between efficiency and equity. AI offers remarkable capabilities – the ability to synthesize information, generate content, answer questions, and assist with countless tasks. These efficiencies create genuine value for users and could theoretically benefit society broadly.

However, these efficiencies come at a cost, and that cost has been disproportionately borne by content creators. While AI companies have raised billions in investment and achieved enormous valuations, the journalists whose work trained these systems have received nothing. The efficiency gains accrue to technology companies and their investors, while the costs fall on creative workers and the institutions that employ them.

This pattern is not unique to AI. Throughout the digital revolution, technology companies have disrupted existing industries by extracting value from content creators while providing minimal or no compensation. Music streaming services pay fractions of a cent per play. Social media platforms built empires on user-generated content without sharing revenues. Now AI companies are training models on the entire corpus of human creative output and claiming they owe nothing to the creators.

The question society must answer is whether this is acceptable. Do we want to live in a world where AI companies can appropriate creative labor without compensation, even when doing so threatens the survival of journalism and other creative industries? Or do we believe that creators deserve recognition and payment for their work, even in the age of AI?

The lawsuit between The New York Times and OpenAI/Microsoft is ongoing, and its outcome remains uncertain. However, regardless of the legal verdict, the case has already served an important function by forcing a public conversation about the value of creative labor in the AI age.

Several potential paths forward exist. One is licensing: AI companies could negotiate agreements with publishers and other content creators to compensate them for the use of their work in training datasets. Some companies, including OpenAI, have already signed licensing deals with certain publishers. However, these agreements are often limited in scope and don’t address the broader question of whether all creators should receive compensation.

Another approach would be regulatory intervention. Governments could establish clear rules about when and how AI companies can use copyrighted material, potentially requiring licensing or creating compulsory licensing schemes that ensure creators receive payment. The European Union has already taken steps in this direction with its AI Act and copyright directives.

A third possibility is industry self-regulation, with AI companies voluntarily adopting practices that respect creators’ rights and provide fair compensation. However, given the competitive pressures in the AI industry and the evidence from the unsealed documents that companies have proceeded despite internal concerns, voluntary restraint seems unlikely without external pressure.

The revelations in the OpenAI and Microsoft copyright case are more than a legal scandal; they’re a referendum on how society values human creativity and labor in the digital age. The alleged circumvention of paywalls, the massive scale of copyrighted material in training datasets, and the internal warnings that went unheeded all point to a system that has prioritized technological advancement over fair compensation for creators.

Brent Hecht’s characterization of AI training as potentially “the largest theft of labor in human history” may sound hyperbolic, but it captures an important truth. Millions of articles, representing countless hours of reporting, writing, editing, and fact-checking, have allegedly been appropriated without permission or payment. The journalists who produced this work are not asking for charity; they’re asking for basic recognition that their labor has value and deserves compensation.

As society continues to grapple with the implications of artificial intelligence, this case serves as a crucial reminder that technology doesn’t exist in a vacuum. Behind every dataset is human labor. Behind every AI-generated response is the accumulated knowledge and creativity of real people who invested time, skill, and often passion into their work. The challenge ahead is to ensure that as we build powerful new technologies, we don’t abandon the principle that creative labor deserves both recognition and reward. Only by finding a path that honors both innovation and equity can society build an AI future that truly serves everyone, not just the companies deploying the technology.