Grok AI update triggers global scrutiny over non-consensual deepfake images on X
text_fields
Grok, the artificial intelligence chatbot developed by Elon Musk’s xAI, has come under global scrutiny after users began exploiting its image-editing feature to create non-consensual, sexually suggestive images, primarily targeting women.
The controversy follows a December 24 update that enabled Grok to edit any public image posted on X when the chatbot was tagged.
Soon after the feature was rolled out, users started requesting edits that altered women’s clothing, appearance, and context in sexually explicit ways without consent. Women with public accounts were particularly affected, with many edits generated by anonymous or burner accounts.
A Bloomberg report cited research by Genevieve Oh, a social media and deepfake researcher, who found that roughly 6,700 sexually suggestive images were being generated by Grok every hour during a 24-hour analysis period. Users also raised alarms about instances involving images of minors, some of which were removed quickly.
Grok publicly admitted and apologised for at least one such incident.
The scale of misuse has prompted action from authorities across multiple regions.
On January 2, India’s Ministry of Electronics and Information Technology directed X to conduct a comprehensive technical, procedural, and governance-level review of Grok’s operations.
European Commission spokesperson Thomas Regnier said the issue was being examined seriously, calling the activity illegal and unacceptable.
Malaysia’s Communications and Multimedia Commission urged platforms to implement safeguards aligned with national laws, particularly for AI-powered image tools.
Musk responded on X by stating that users generating illegal content through Grok would face the same consequences as those uploading illegal material.
However, neither X nor xAI has clarified what safeguards were absent or how violations would be enforced. As of the latest reports, Grok continues to comply with some requests for sexually inappropriate image edits.