WikiHow accuses OpenAI of using over 11,000 articles to train AI models

WikiHow, the world's largest collaborative 'how-to' guide website, has filed a lawsuit against OpenAI in a U.S. federal court in Manhattan. The lawsuit alleges that OpenAI unlawfully scraped more than 11,000 tutorial articles from WikiHow to train its ChatGPT and GPT large language models. The company claims that OpenAI's actions have infringed on at least 1,200 registered copyrights. According to the lawsuit, the stolen content spans a wide range of topics, from everyday tasks to professional skills. OpenAI is accused of using this material to generate similar tutorial content at a much lower cost than producing original guides, thereby reducing WikiHow's revenue and potentially diminishing its motivation to continue creating content. WikiHow is seeking court orders to stop OpenAI from further infringing on its copyrights and is requesting damages, though the exact amount has not been disclosed. OpenAI responded by stating that its models are trained using publicly available data under the principle of fair use. Representatives from WikiHow have not yet commented on the matter.