General outcry. A few days ago, Internet users tiquated new general conditions for the use of Wetransfer, the file sharing service. Modified on June 23, they mentioned since Internet users agreed to let the site ” [leur] Content for exploitation, development, marketing and improvement of the service or new technologies or new services, including to improve the performance of automatic learning models that improve our content moderation process, in accordance with confidentiality and cookies policy. “
It didn’t take any more so that people shout in scandal. This formulation suggests that all the files pass through the site come to lead an AI. Internet users were concerned in particular to know that very professional or personal files could be read, and that their data could be regurgitated in the future by a conversational agent.
An authorized practice, but which does not delight everyone
The practice is however authorized. “There is no French law on the learning of major models,” recalls Alain Bensoussan, lawyer expert in digital law. On the European side, the AI regulation came into force in August 2026, and has no obligation if not transparency. Users are therefore linked by the conditions of use of software. If the shared data is personal, it is up to the transmitter to pay attention to the security of exchanges is ensured. The thorny question of copyright and intellectual property of training data, in particular for image producers, is the subject of a legal vacuum.
Despite the reluctance of part of the public for AI, Alain Bensoussan still notes a nuance. “There is an advantage in having free tools, points out the lawyer. Artificial intelligence allows for example access for all to a quantity of information on past and present civilizations. Is it better a free chatgpt that trains on data or a paid tool for everyone? »»
In any case, facing Bad Buzz, Wetransfer was however forced to return to the situation, specifying that this clause only concerned the moderation tools of the site. “We do not use automatic learning or any other form of artificial intelligence to treat shared content via Wetransfer, and we do not sell content or data to third parties either,” added a company spokesperson to the BBC. The formulation of the conditions of use of the service has since been modified, erasing the mention of “automatic learning models”.
