Japan opens probe into Elon Musk's Grok AI after sexualized images

Japan's Cabinet Office said it had opened a formal investigation into Grok, the artificial intelligence service developed by xAI and integrated into Elon Musk's X platform, after researchers and users demonstrated that the system could generate sexualized and explicit images of real people. The move marked Tokyo's entry into a widening international backlash over the chatbot's image-generation and editing capabilities.

The Cabinet Office formally asked X Corp. to implement immediate improvements to Grok's safeguards and to curb the output of sexually altered or otherwise inappropriate images. Economic Security Minister Kimi Onoda, who also oversees the government's AI strategy, said X had not yet responded to the request and warned that the government "plans to promptly examine all possible options, including legal measures," if X did not improve protections.

xAI responded in public statements in the days before Japan's announcement, saying it had implemented technical tweaks intended to prevent Grok users from editing "images of real people in revealing clothing such as bikinis." The company also said it had blocked users based on their location from generating images of people in revealing clothing "where it's illegal." xAI's safety team communicated that the restriction applied to all users, "including paid subscribers." Japanese officials said those steps were insufficient at the time the Cabinet Office issued its request.

The probe deepens a pattern of international scrutiny that began after demonstrations showed how Grok could be prompted to produce sexualized images of women and children and to create images that raise privacy, consent and intellectual property concerns. Britain and Canada have said they are advancing their own inquiries, while Malaysia and Indonesia temporarily blocked access to Grok after it was used to create explicit images. India reported that X removed thousands of posts and hundreds of user accounts in response to complaints.

U.S. state authorities have also moved. California's attorney general launched an investigation into xAI over sexually explicit material and allegations that such imagery was used for harassment online. Britain's communications regulator opened an inquiry into whether X complied with domestic law. France's children's commissioner referred Grok-generated images to prosecutors and national media and content regulators for possible action.

Regulators and privacy advocates say the case highlights limits in current content-moderation tools and the difficulty of policing AI systems that can both generate and edit realistic imagery. Geoblocking based on user location is a partial mitigation, but experts caution it can be bypassed and does not address the core risks of misuse or the creation of nonconsensual imagery that can be distributed widely.

Tokyo's probe will focus on whether X's technical fixes and policy controls meet Japanese legal and safety expectations and on whether the company provides an adequate timetable for concrete changes. Officials also plan to seek clarity on which jurisdictions xAI has geoblocked and the technical scope of those restrictions.

The outcome in Japan could influence how other governments press platforms to design and enforce guardrails for generative AI. For now, regulators and platform owners face a fast-moving technical landscape and public pressure to prevent harms while the underlying capabilities of image-generating models continue to evolve.