Key Points
- Elon Musk’s AI chatbot, Grok, generated inappropriate images of minors.
- Grok’s official account acknowledged “lapses in safeguards” and said they are “urgently fixing” the issue.
- When asked for comment, xAI replied with the message “Legacy Media Lies.”
- Governments in France and India have launched investigations into the matter.
Elon Musk’s AI chatbot, Grok, is in hot water after it was found to be creating inappropriate images of children. On Friday, the chatbot’s official X account acknowledged that “lapses in safeguards” had allowed users to generate “images depicting minors in minimal clothing.” The company said it is “urgently fixing” the problem.
The issue came to light when users on X started sharing screenshots of Grok’s public media tab. The images, which users said were altered by the AI after they uploaded photos and prompted the bot to change them, quickly drew widespread condemnation.
Grok’s statement acknowledged the problem and said that while they have safeguards in place, they are working on “improvements to block such requests entirely.”
When Reuters reached out to Musk’s company, xAI, for a comment, they replied with the message “Legacy Media Lies.” However, governments around the world are clearly taking the problem seriously.
Ministers in France have already reported the “sexual and sexist” content to prosecutors, calling it “manifestly illegal.” They are also asking regulators to see if it violates the European Union’s Digital Services Act.
Meanwhile, India’s IT ministry sent a letter to X’s local office, stating that the platform failed to prevent misuse of Grok. They have ordered X to submit a report on the actions they’ve taken within three days.
This is a major setback for Musk’s AI venture, which is competing with established players such as OpenAI and Google. While all AI companies have struggled with content moderation, this incident involving minors is particularly damaging. Grok has said that it is prioritizing improvements, but it also admitted that “no system is 100% foolproof.”
