WikiHow sues OpenAI for copyright infringement over AI training
WikiHow, Inc. filed a copyright infringement lawsuit against OpenAI, Inc. on August 21, 2026, in the U.S. District Court for the Southern District of New York, in Manhattan federal court Reuters. The complaint alleges that OpenAI scraped more than 11,000 of WikiHow's how-to articles to train ChatGPT and that the lawsuit claims infringement of at least 1,200 registered copyrights Reuters.
Lawsuit Details and Claims
WikiHow seeks both monetary damages and an injunction to stop further infringement Reuters. The filing represents a concrete enforcement action by a content platform against a major AI company over unauthorized training data use—a core flashpoint in the emerging agent economy where large language model training depends on massive corpuses of human-created material.
The lawsuit joins a broader wave of copyright disputes targeting foundational model developers. Multiple content publishers and creators have challenged AI companies on similar grounds, but WikiHow's case is notable for its specificity: the complaint identifies a precise volume of articles (11,000+) and number of registered copyrights (1,200+) allegedly infringed. This granularity provides the court with concrete metrics to evaluate damages and injunction scope.
Real-World Impact on Agent Training Infrastructure
The case directly implicates how LLM-backed agents are built and trained. ChatGPT, which powers many autonomous agent implementations in enterprise and consumer settings, was trained on scraped content that WikiHow contends was taken without license or consent. If WikiHow prevails or secures a preliminary injunction, it could establish precedent for content platforms to restrict AI training access or demand licensing fees—potentially reshaping the cost structure and legal risk profile of building new agent systems.
The filing also signals that content creators are willing to litigate rather than settle quietly. Unlike some earlier copyright disputes that resulted in sealed settlements, WikiHow chose a public federal court venue, exposing OpenAI's training practices to discovery and public scrutiny.
What Comes Next
The case is now in the early stages of federal litigation. No trial date has been set, and OpenAI has not yet filed a response to the complaint. Outcomes could range from a settlement with licensing agreements to a ruling that forces modification of existing models or establishes liability for past training use. The precedent will likely inform how other platforms and publications approach AI companies seeking access to their content.