Jay Edelson has spent the better part of two decades making the technology industry pay — sometimes billions of dollars at a time. Now, with artificial intelligence companies vacuuming up the world’s data to train their models, the Chicago-based plaintiffs’ attorney sees his biggest opening yet. And he’s not being subtle about it.
Edelson, the founder of Edelson PC, has emerged as one of the most consequential legal figures in the fight over how AI companies acquire and use training data. His firm has filed a barrage of lawsuits against major AI players, including OpenAI, Meta, and Google, alleging they scraped copyrighted content and personal data without permission to build the large language models now generating tens of billions in revenue. The stakes are staggering. If Edelson and attorneys like him prevail, the financial foundations of the generative AI boom could be fundamentally rewritten.
This is not new territory for him. Not exactly.
As The Information detailed in a recent profile, Edelson first made his name by going after Facebook over its use of facial recognition technology. That case, which alleged Facebook violated Illinois’ Biometric Information Privacy Act by scanning users’ faces without consent, resulted in a $650 million settlement in 2021 — one of the largest privacy settlements in U.S. history. It established Edelson as someone who could extract enormous sums from companies that most attorneys were too intimidated or too conflicted to challenge.
His approach is confrontational by design. Edelson has publicly criticized what he sees as a cozy relationship between the plaintiffs’ bar and Big Tech, where class-action attorneys negotiate settlements that deliver minimal payouts to consumers while generating lucrative fees for the lawyers themselves. He’s called these arrangements “sellout settlements” and has made a career out of objecting to them in court, often successfully. It’s a posture that has earned him enemies on both sides of the aisle — defense attorneys who view him as a publicity-seeking agitator and fellow plaintiffs’ lawyers who resent being called out.
But it’s also a posture that has earned results.
The AI cases represent a significant escalation. Unlike earlier privacy disputes, which turned on relatively narrow statutory violations, the training-data lawsuits raise sweeping questions about intellectual property, fair use, and the limits of corporate data collection. Edelson’s firm has argued that companies like OpenAI ingested vast libraries of copyrighted books, articles, and other creative works to build models like ChatGPT — without compensating or even notifying the authors. The firm represents individual plaintiffs, including authors and content creators, who say their work was consumed by AI systems that now compete directly with them.
OpenAI has countered that its use of publicly available data constitutes fair use under copyright law, a defense that will likely be tested in courtrooms for years. The company and its allies argue that restricting access to training data would stifle innovation and hand advantages to foreign competitors, particularly Chinese AI firms that face fewer legal constraints. It’s an argument that carries real weight in Washington, where policymakers are reluctant to impose rules that might slow American AI development.
Edelson isn’t buying it. As he told The Information, the tech industry’s innovation arguments are a familiar deflection — the same kind of rhetoric Facebook deployed when it resisted accountability for its facial recognition practices. In his view, the pattern is consistent: move fast, take what you need, and argue after the fact that the public benefit justifies the taking.
The legal terrain is shifting rapidly. In recent months, courts have begun issuing early rulings that provide partial signals about how these cases might unfold. A federal judge in the Northern District of California allowed several claims against OpenAI and Microsoft to proceed, rejecting motions to dismiss that argued the plaintiffs lacked standing. Meanwhile, separate litigation brought by The New York Times against OpenAI — one of the highest-profile training-data cases — continues to advance, with the newspaper arguing that OpenAI’s models can reproduce near-verbatim passages of its reporting.
The proliferation of lawsuits has created a dense web of overlapping claims. Authors, visual artists, musicians, software developers, and news organizations have all filed suits alleging unauthorized use of their work. Some of these cases have been consolidated in multidistrict litigation proceedings, while others proceed independently. Edelson’s firm is involved in several threads simultaneously, positioning itself as a central node in what has become one of the most complex areas of technology litigation in decades.
What makes Edelson particularly dangerous to defendants, according to attorneys who have faced him, is his willingness to reject settlements that other firms would eagerly accept. The economics of class-action litigation typically push toward resolution — trials are expensive, uncertain, and time-consuming. Most plaintiffs’ firms operate on contingency and need settlements to keep the lights on. Edelson has built a practice model that allows him to be more patient, more aggressive, and more selective about the deals he’ll accept.
That stubbornness has occasionally backfired. Edelson has faced criticism from judges who found his objections to proposed settlements disruptive, and some legal observers argue that his combative style sometimes prioritizes publicity over pragmatism. He’s been sanctioned in at least one case, and his firm’s aggressive tactics have drawn scrutiny from courts that questioned whether certain filings were brought in good faith.
None of that appears to have slowed him down.
The broader context matters here. The AI training-data fight is unfolding against a backdrop of intense political and regulatory uncertainty. The Trump administration has signaled a permissive approach to AI development, rolling back Biden-era executive orders on AI safety and resisting calls for comprehensive federal AI legislation. That regulatory vacuum has elevated the importance of private litigation as the primary mechanism for establishing guardrails around AI companies’ behavior.
Congressional action remains unlikely in the near term. While several bills addressing AI and copyright have been introduced — including proposals that would require AI companies to disclose their training data sources and obtain licenses for copyrighted material — none have gained sufficient momentum to advance through committee. The lobbying firepower arrayed against such legislation is formidable. OpenAI, Google, Meta, and Microsoft have collectively spent hundreds of millions on lobbying and political contributions, and their argument that heavy-handed regulation would cripple American competitiveness resonates with lawmakers in both parties.
So the courts become the arena. And Edelson is treating it as such.
His firm’s strategy in the AI cases reflects lessons learned from the Facebook biometric litigation. In that fight, Edelson’s team identified a specific state statute — Illinois’ BIPA — that provided clear statutory damages and a private right of action, giving plaintiffs powerful tools that federal privacy law generally lacks. The result was a case that was difficult for Facebook to dismiss and expensive to lose. Edelson is now looking for similar pressure points in the AI context, combining federal copyright claims with state-level privacy and consumer protection statutes that may provide additional avenues for relief.
The financial implications for AI companies are hard to overstate. If courts determine that training on copyrighted data without permission constitutes infringement, the potential damages could reach into the tens of billions — enough to threaten the economics of companies that have raised capital at valuations predicated on the assumption that their training practices are legal. OpenAI, valued at over $150 billion in its most recent funding round, has built its entire product line on models trained with data whose provenance is now being litigated. A ruling that requires retroactive licensing or damages could force a fundamental restructuring of how AI companies operate.
Some companies are already adapting. Adobe, for example, has trained its Firefly image generation model exclusively on licensed content and public domain material, positioning itself as a “safe” alternative for enterprise customers concerned about legal exposure. Shutterstock and Getty Images have struck licensing deals with AI companies, creating revenue streams from their content libraries. These arrangements suggest that a market-based solution is possible — but only if courts create sufficient legal pressure to force AI companies to the negotiating table.
That’s exactly what Edelson is trying to do.
The coming months will be critical. Several of the major AI training-data cases are moving toward discovery phases, where plaintiffs will seek internal documents and communications from AI companies about their data collection practices. These disclosures could prove explosive. In the Facebook biometric case, internal documents revealed that the company was aware of potential legal risks from its facial recognition technology but proceeded anyway — evidence that proved devastating in settlement negotiations. Edelson and other plaintiffs’ attorneys are betting that similar documents exist inside OpenAI, Google, and Meta.
The AI companies are fighting hard to prevent that. Motions to limit discovery, seal documents, and narrow the scope of claims are being filed aggressively across multiple jurisdictions. Defense attorneys argue that forcing disclosure of proprietary training methodologies would compromise trade secrets and competitive advantages. It’s a tension that courts will have to resolve on a case-by-case basis, and the outcomes will shape the trajectory of the entire litigation wave.
For Edelson, the AI fight is personal in a way that goes beyond professional ambition. He has spoken publicly about what he sees as a moral dimension to the dispute — the idea that creative workers are being exploited by corporations that profit from their labor without compensation. It’s a framing that resonates in an era of widespread anxiety about AI-driven job displacement, and it gives his cases a populist energy that purely commercial copyright disputes often lack.
Whether that energy translates into courtroom victories remains to be seen. The fair use doctrine is notoriously elastic, and courts have historically been reluctant to issue rulings that could be perceived as inhibiting technological progress. The AI companies have deep pockets, elite legal teams, and powerful political allies. They won’t go quietly.
But Jay Edelson has heard that before. He heard it from Facebook. He heard it from Google. He heard it from every company that told him the law was on their side and the settlement would be on their terms. He kept pushing anyway. And more often than not, the check arrived.
The question now is whether the AI industry’s check will be the biggest one yet.
WebProNews is an iEntry Publication