The New York Times sued Microsoft and OpenAI yesterday for mass copyright infringement, estimating damages to be worth billions of dollars. The Times, the first major media company to file a lawsuit against Microsoft and OpenAI, accuses them of using its articles to train artificial intelligence chatbots without permission.
Large language models (see overview) scrape information available online to help AI platforms generate images or human-like text. The Times said at least 16 million unique records of content from its publications were used in a dataset underpinning OpenAI's ChatGPT-3. The Times also said such platforms could reduce traffic to its site and decrease revenue from advertising and subscriptions. Among the examples in its filing (read here), the Times showed both ChatGPT and Microsoft's Browse With Bing reproduced nearly verbatim information from the Times (starting p. 29) and how ChatGPT can bypass its paywall when asked to do so (p. 33).
OpenAI also faces a similar lawsuit from novelists, including George RR Martin and John Grisham.