WikiHow sues OpenAI for scraping over 11,000 articles without permission

WikiHow has filed a copyright infringement lawsuit against OpenAI, claiming the AI company scraped more than 11,000 of its instructional articles to train GPT models, including ChatGPT, without ever asking permission. The complaint, filed August 21, 2026, in the US District Court for the Southern District of New York, lands OpenAI in familiar territory: it is now the defendant in at least 24 copyright cases brought in US courts. The suit does not name a specific damages figure, but it seeks both monetary compensation and injunctive relief, which could include a court order forcing OpenAI to stop using the scraped content in future training runs. What WikiHow is actually claiming The core argument is straightforward. WikiHow says its team spent years producing detailed, illustrated how-to guides, and OpenAI used more than 11,000 of those articles as raw training material without a license, a payment, or even a heads-up. The complaint asserts that at least 1,200 of those articles carry registered copyrights, which matters because registered works unlock statutory damages in US court, a much sharper financial weapon than actual damages alone. WikiHow’s second argument is economic. Cha...