OpenAI, the company behind popular artificial intelligence (AI) interface ChatGPT, has been sued in the United States by two authors for copyright infringement.
The authors, Paul Tremblay and Mona Awad, have filed a class action in a federal court in San Francisco alleging that the company made non-consensual, non-credited and non-remunerated use of its works to "train" ChatGPT. According to the lawsuit, ChatGPT extracts data copied from thousands of books without any type of permission, in order to generate its responses with a surprisingly human style.
The authors have pointed out that ChatGPT is capable of preparing detailed summaries of their works, which would prove that the AI has had direct access to them, even if they are not freely available on the Internet.
This is not the first lawsuit of these characteristics that has been filed in the US. Visual artists have already initiated actions against Midjourney, DevianArt and Stable Diffusion, AIs that generate illustrations based on suggestions or verbal "prompts" from users.
The AI "learning" system requires the processing of millions of sources of text and images. Some of these sources are in the public domain but the vast majority, like the novels of the authors of this lawsuit, are protected by copyright. This and other similar lawsuits will be very interesting to learn how AI companies actually work, as well as to develop the currently non-existend legal framework on the matter, and which seems to endanger many cultural industries.