Grok: Preventing “Undressing” of Real People Online
(San Francisco) Le réseau social X d’Elon Musk a annoncé mercredi avoir déployé des mesures pour « empêcher » son outil d’intelligence artificielle (IA) Grok de « déshabiller » des « personnes réelles », mais doit encore convaincre les autorités de plusieurs pays qui ont ouvert des enquêtes.
Publié hier à
« Nous avons mis en place des mesures technologiques pour empêcher le compte Grok de permettre l’édition d’images de personnes réelles dans des tenues révélatrices, comme des bikinis », indique X dans un message publ
As of January 9th, image generation with Grok, the AI assistant developed by xAI, remains limited to paying subscribers, X confirms.
Earlier in January,the platform stated it was “taking action against illegal content […] by removing it,permanently suspending accounts,and collaborating with local authorities.”
However, it remained silent afterward despite growing protests and threats of official sanctions.
Some users utilized Grok to post on X or reply to messages containing photos of women, accompanying them with instructions like “Put her in a bikini” to generate hyperrealistic montages (“deepfakes”).
California Inquiry
According to an analysis last week by the NGO AI Forensics, conducted on over 20,000 images generated by Grok, more than half depicted scantily clad people, with 81% being women and 2% appearing to be minors.
The scandal led to the suspension of Grok in