OpenAI, the company behind ChatGPT, is sued. A class action led by a law firm believes that artificial intelligence had no right to train on open access publications, such as Wikipedia articles or comments on social networks.
ChatGPT is definitely one of the most popular and powerful artificial intelligences around. It must be said that the chatbot developed by OpenAI trained on billions of data to reach this level and its impressive results. What is less known is exactly what data OpenAI used to improve its AI — and whether it had the right to use it. To answer this thorny question, a complaint was filed.
A California law firm has launched a class action lawsuit against OpenAI, The Washington Post learned on June 28, 2023. The firm believes the company massively violated the copyrights and privacy of countless people by dragging ChatGPT on data retrieved from the Internet, without asking permission, writes the Washington Post.
Was ChatGPT allowed to train on this data?
The law firm says it wants to help ” people whose information was stolen and misused for commercial purposes to create this very powerful technology Director Ryan Clarkson told The Washington Post. For him, there are many comments on social networks, blog posts, Wikipedia articles and many other contents that would be affected.
ChatGPT has indeed trained on huge amounts of open access data — usually without worrying about copyright issues. This particularity was already making many Internet users cringe, annoyed at having been used without their consent and without financial compensation to train a super powerful tool – and sold at a very high price.
” All this information is used on a large scale, while it was never intended to be used by an artificial intelligence “said Ryan Clarkson to the Washington Post. The lawyer hopes to obtain compensation for his clients, and above all, to ensure that safeguards are put in place regarding the use of data by AIs. There is indeed still a legal vagueness around the use of publications in free access on the Internet by third parties – and it is exactly on this aspect that the American justice will have to decide.
ChatGPT is not alone in being affected by a copyright infringement complaint. The image-generating AI Stable Diffusion has been sued by the photo agency Getty Image, which believes that the AI trained illegally on its photos – to the point where Stable Diffusion manages to reproduce the famous logo of the agency. As for Midjourney, it is also in legal turmoil: several artists have filed a complaint against the company, accused of having trained its AI with their works, without ever having asked for their agreement.
Subscribe for free to Artificials, our AI newsletter, designed by AIs, verified by Numerama!