In a recent analysis of Reddit’s artificial-intelligence licensing strategy, the growing value of data, and Reddit’s upcoming Initial Public Offering (“IPO”), Barron’s interviewed KellDann’s Kirk Sigmon for insight into Reddit’s lawsuit against Anthropic. Kirk’s comments also appeared on various other Wall Street Journal-affiliated publications, like Law.com.
Reddit alleges that Anthropic scraped data from its platform to train Anthropic’s AI models (like Claude) in violation of Reddit’s User Agreement. Kirk highlighted why the legal theory matters: unlike many prominent AI training-data disputes, Reddit’s lawsuit focuses on user-agreement violations rather than copyright infringement. That distinction, Kirk explained, will limit Anthropic’s ability to rely on fair use. Reddit’s complaint can be read here.
The Reddit v. Anthropic case illustrates a broader issue for AI developers and data-rich platforms: legal exposure associated with training data may turn not only on copyright law, but also on how data is accessed, what contractual restrictions govern its use, and whether appropriate licensing arrangements are in place. Concepts like “clickwrap” and “browserwrap” agreements are, in some ways, becoming more important than ever.