Reddit: Where you can simultaneously learn how to fix your car, ruin your sleep schedule, and discover that your most embarrassing moment is actually pretty tame. With over 100 million daily users and hundreds of thousands of communities, the platform is no longer just the safe haven for introverts who want to discuss Linux distros and obscure internet memes.
But when you get that big, you become coveted, and the battle for control of the world’s treasure trove of unfiltered human conversation has officially begun. The culprit? Greedy AI companies.
A Digital Gold Rush
Reddit has long prided itself on being the steward of one of the largest datasets of natural human language discussion in existence. But as artificial intelligence technologies have exploded in popularity and capability, the value of Reddit’s vast corpus has grown exponentially—especially as a potential source for training large language models (LLMs).
Reddit’s content, though publicly accessible, is governed by a User Agreement and Privacy Policy that purport to restrict commercial exploitation and protect user privacy. According to Reddit, these guardrails are essential not only to safeguard its business interests but also to honor the rights and expectations of its users. As the company would go on to explain in its complaint filed in California state court, “[s]imply because Reddit content is publicly available does not mean others are free to exploit it for their own commercial advantage without permission or compensation.”
Anthropic Crosses a Line?
Anthropic, a company that spun out from OpenAI, has positioned itself as a champion of ethical AI development. Its flagship product, Claude, is an AI chatbot that quickly became one of the most popular alternatives to OpenAI’s ChatGPT. Anthropic touts its commitment to “honesty” and “high trust,” claiming that it does not intend to train its models on personal data or violate privacy norms.
But Reddit’s lawsuit paints a very different picture. The complaint alleges that as far back as December 2021, Anthropic was already “training Claude on Reddit users’ posts” without authorization and in direct violation of Reddit’s User Agreement. The complaint quotes Anthropic researchers, including CEO Dario Amodei, as saying that “Training [AI models] on large public preference modeling data sourced from ... Reddit comments ... significantly improves sample efficiency when subsequently finetuning on small preference modeling datasets.”
Reddit further alleged that Anthropic had continued scraping Reddit content even after the AI company publicly claimed to have blocked its bots from accessing the platform. According to Reddit’s audit logs, Anthropic’s bots accessed (or at least attempted to access) content from Reddit more than 100,000 times after July 2024.
The Stakes
Reddit’s core grievance centers on what it calls the “unauthorized commercial use” of its content, use that allegedly harms both Reddit and its users. Unlike other tech giants such as OpenAI and Google (who have entered into formal licensing agreements with Reddit), Anthropic allegedly refused to engage in licensing negotiations or comply with Reddit’s privacy-protective guardrails.
Reddit claims these licensing agreements are not mere formalities; they are designed to ensure compliance with deletion requests, restrict access to sensitive content, and provide transparency about who is accessing user data and why. Without such agreements, Reddit argues, users lose meaningful control over their data and privacy.
The complaint lays out five causes of action: breach of contract (violation of the User Agreement), unjust enrichment (profiting off Reddit content without compensation), trespass to chattels (interference with Reddit’s servers and infrastructure), tortious interference with contract (undermining Reddit’s obligations to its users), and unfair competition under California law.
Anthropic Claims It’s Really About Copyright
Anthropic wasted no time firing back, arguing that Reddit’s lawsuit was fundamentally a copyright dispute dressed up in state law claims. In its formal legal response, Anthropic asserted that Reddit’s allegations were “quintessential copyright claim[s],” but that Reddit did not own the copyrights to user-generated content posted on its platform. Reddit might have “artfully plead its copyright-like claims in contract and tort,” Anthropic argued, but these labels “do not and cannot change the gravamen of Reddit’s suit or the proper forum for it.”
Anthropic argued that because the dispute implicated federal copyright law, it should be heard in federal court, not state court, where Reddit had filed the complaint. The company cited prior decisions that found complete preemption where claims did not implicate rights qualitatively different from those protected by the Copyright Act. A spokesperson for Anthropic described Reddit’s approach as an attempt “to avoid a pure copyright fight because those claims are harder to prove.” Intellectual property attorney Vivek Jayaram agreed: “State law gives them more levers to protect their data.”
Reddit rejected Anthropic’s characterization outright. A spokesperson for Reddit insisted that the “case isn’t about copyright at all” and that Anthropic was mischaracterizing the complaint “to distract from their misconduct and the point of this lawsuit.” Reddit maintained that each of its claims contained an “extra element” not required for a copyright claim — such as breach of contract or interference with user privacy — and therefore were not preempted by federal law.
Reactions from the Rest of the World
The lawsuit has drawn sharp reactions from across the tech industry. Adam Eisgrau, senior director at the Chamber of Progress (a Big Tech-backed advocacy group), sided with Anthropic. “Reddit’s transparent attempt to sue Anthropic for copyright infringement by dressing up that exclusively federal action as a brace of state claims remains an inartful dodge that’s wasting the court’s time and resources,” he said.
Others see broader implications for how online platforms can control—or monetize—their public data in an era where AI training sets are worth billions. If courts side with Reddit, platforms may gain new leverage over how their user-generated content is used by AI companies; if Anthropic prevails, scraping public data could remain fair game for model training unless Congress steps in.
What Happens Next?
At present, much remains unresolved, and the parties continue to spar over jurisdiction. What is clear is that this case will test lots of new legal boundaries beyond just copyright law. If it doesn’t settle, it could answer some important new questions about contractual rights over user-generated data, privacy expectations in social media platforms, and the commercial realities of AI development. The outcome could set precedent for how online communities interact with artificial intelligence for years to come.
Related Resources:
- Thomson Reuters Wins Possible Landmark Copyright Case Against AI Company (FindLaw's Federal Courts)
- AI Company Anthropic Defends Against (Some) Copyright Claims with Fair Use Doctrine (FindLaw's Federal Courts)
- Artificial Intelligence Company Pays $1.5 Billion in Settlement, but Did It Really Lose? (FindLaw's Courtside)
