South Korea Launches Investigation Into Grok Over Deepfake Image Scandal

Global Outcry Over AI-Generated Explicit Content

South Korea's Personal Information Protection Commission has launched a preliminary investigation into Grok, the AI chatbot developed by Elon Musk's xAI, following disturbing reports that the tool has been used to generate millions of sexually explicit deepfake images.

The investigation comes after the Center for Countering Digital Hate estimated that Grok generated approximately three million sexualised images over an eleven-day period between late December and early January, averaging roughly one hundred and ninety images per minute. Among these, an estimated twenty-three thousand images depicted minors.

Regulatory Pressure Mounts

South Korea's Media and Communications Commission has demanded that X implement stronger youth protection measures, giving the platform two weeks to submit a plan preventing the generation of illegal content and restricting minors' access. Failure to comply could result in administrative fines.

Under South Korean law, generating or altering sexual images of identifiable individuals without consent may constitute unlawful handling of personal data, making this a serious regulatory matter.

International Coalition Forms

The investigation places South Korea alongside a growing list of nations taking action. Indonesia and Malaysia have blocked access to Grok entirely, whilst the Philippines has restricted access pending content removal. In the United States, California's Attorney General launched an investigation and issued a cease-and-desist order, whilst thirty-five state attorneys general signed a bipartisan letter demanding immediate action.

The UK's Ofcom has opened an Online Safety Act compliance investigation, Canada's Privacy Commissioner is examining Grok specifically, and the European Commission has ordered X to preserve all internal documents related to the chatbot.

Company Response Under Scrutiny

xAI has announced restrictions limiting image generation to paying subscribers and blocking the editing of real people. However, regulators in multiple countries have described these responses as insufficient, noting they rely heavily on user reporting rather than addressing fundamental design flaws in the AI system.