WikiHow, the how-to site that has spent two decades publishing step-by-step guides on everything from changing a tire to writing a cover letter, has sued OpenAI.
According to MediaNama, the lawsuit alleges that OpenAI scraped WikiHow's copyrighted articles to train its AI models without permission and without compensating the publisher. The coverage was picked up and syndicated through Google News and Bing News aggregators, both of which carried MediaNama's report of the filing.
The available reporting is thin on specifics. Neither source names the court, the damages sought, the specific OpenAI models at issue, nor does either include a response from OpenAI. What is established is the core allegation: unsanctioned use of WikiHow's content as AI training material.
The shape of the complaint will be familiar to anyone following the AI industry. Publishers argue that their archives — expensive to produce, and in WikiHow's case built substantially through community contribution and editing — were vacuumed up to build commercial AI products that can now answer the same questions the original articles were written to answer. AI companies have generally argued that training on publicly accessible material is permissible.
WikiHow is a particularly pointed plaintiff. Its entire business is instructional content, exactly the kind of practical, procedural knowledge that chatbots are frequently asked to supply. If a user asks an assistant how to unclog a drain and gets a serviceable answer, they never visit the site that taught the model to give it.
This matters because the outcome of cases like this one will determine whether the web's how-to and reference publishers get paid for the material that makes AI assistants useful — or whether they simply lose the traffic that funded it.