Two engineers believe that OpenAI, the company that developed ChatGPT, and Microsoft, which has invested billions in this generative AI, illegally collected billions of data from social networks and other web platforms to train their system. The two scientists are at the origin of a new collective procedure in the United States. They could be joined by millions of Internet users.
Hundreds of millions of Internet users looted? This is what this new legal action targeting OpenAI and Microsoft suggests. As last June, the start-up behind ChatGPT and the software giant – which has invested billions in the company – are both targeted by a class action in the United States. They are accused, according to the complaint filed on September 5 in San Francisco federal court, of having stolen and used personal data of hundreds of millions of Internet users to train ChatGPT.
At the origin of this class action, there are two engineers, who wished to remain anonymous. This type of procedure makes it possible to attack, in the name of thousands of people and in one go, a company – with the result of damages requested, the amount of which is not specified here. The two scientists believe that the two companies violated several American privacy laws by developing ChatGPT, the most popular chatbot of the moment, capable of generating text and code.
Data recovered “largely in secret”
This type of generative AI system requires immense amounts of data to be successful. But these data would have been, according to the complaint, recovered “ largely in secret ”, without authorization, and in violation of the terms of use of many websites and the rights of individuals to refuse such collection. Among these data are listed in the complaint “ private information and conversations » from messaging between individuals or professionals, as well as numerous “ intercepted information such as user locations, musical tastes, financial or medical data “.
Another complaint: engineers also fear that companies have incorporated their “ skills and expertise » in products that could “ one day train [leur] professional obsolescence “.
Gold ” without this unprecedented theft of private and copyrighted information belonging to real people “, it is written in the complaint, OpenAI “ wouldn’t be the multi-billion dollar company today “.
Also read: “Stop stealing our books, and pay us”: in the United States, the bronca of writers in the face of AI manufacturers
Objective: “make “Big AI” responsible for their massive theft of personal information”
This is the second time that collective action has been taken against OpenAI’s generative AI tool for theft of data from Internet users. Already in June, a similar procedure targeted the developer of ChatGPT. Ryan Clarkson, one of the lawyers from the Clarkson firm in charge of this first legal action, was delighted with this news. The lawyer explained in a press release on Wednesday September 5 that the objective was to “ holding “Big AI” accountable for their massive theft of personal information and violations of privacy, property and consumer rights », Reports Reuters.
Also read: OpenAI sued for using your comments and blog posts
Other actions, this time for theft of data protected by copyright, are also underway. They target Stabile Diffusion, Midjourney and DeviantArt, who are accused of having collected billions of images in order to “ learning » of these AIs, including creations protected by copyright.
Also read: A great “digital plunder”: when generative AI challenges copyright
The engineers have asked the court for an unspecified amount of damages to be paid to any member of this proceeding – which could include anyone whose information was recovered. They also require companies to implement safeguards against the misuse of private data. Questioned by our colleagues, neither OpenAI nor Microsoft wished to comment on this matter.
Source :
Complaint filed on September 5, 2023