Elon Musk pulls the plug: Grok is being stopped from using deepfakes.

The AI ​​chatbot Grok is currently under heavy fire because users have misused the tool to create fake pornographic images. For a long time, X CEO Elon Musk ignored the accusations or dismissed them as censorship attempts, but the global pressure has now apparently become too great. 

Several countries, including Malaysia and Indonesia, have already completely blocked access to the AI. Investigations are also underway in Europe and the US, as Grok created sexualized deepfakes of real people – tragically, even minors – without any significant restrictions.

X has now responded with a drastic change to the guidelines for XAI's AI model. The image editing and generation functionality has been significantly restricted. According to an official statement, new technological barriers now prevent Grok from placing photos of real people in sexual contexts. This primarily affects commands that digitally undress people or dress them in revealing outfits, such as bikinis. These restrictions apply immediately to all users, including paying subscribers of the premium service.

To further enhance security, X has placed the entire image function behind a paywall. Only those who pay for a subscription and thus verify their identity via payment data are permitted to use the AI ​​tools. Musk hopes this will better deter perpetrators or, if necessary, hold them accountable. Additionally, the platform is implementing geoblocking. In regions where creating such deepfakes is illegal, X completely disables the corresponding functions for affected users.

Critics, however, doubt whether these measures are sufficient. While direct commands are now blocked, the AI ​​can often still be tricked with creative rewrites. Elon Musk continues to maintain that Grok never intentionally violated any laws and that technical errors are immediately corrected. Whether this will reassure regulators in the EU or California remains to be seen. The case once again demonstrates the dark side of generative AI and the difficulty of effectively preventing abuse on large social media platforms.

Share this post