Asger Risom
Grok has been pulled back from the brink after weeks of mounting international pressure, as regulators and governments moved to curb the spread of sexualized AI images generated without consent.
X has moved to restrict its AI chatbot Grok from generating sexualized images of real people, following intense global backlash and regulatory scrutiny over the misuse of the tool.
The social-media platform confirmed it has implemented new technical safeguards to prevent Grok from editing or creating images of real individuals in revealing clothing, including bikinis and underwear. The restriction applies to all users, including those paying for premium access, and marks a sharp reversal after days of criticism that earlier measures did not go far enough.
The move came just hours after California Attorney General Rob Bonta announced an investigation into the creation of sexualized AI deepfakes, including images involving minors, allegedly generated using Grok.
Regulators move in as pressure mounts
Authorities said they had received a surge of reports alleging that Grok users were taking images of women and children found online and using the AI model to generate non-consensual sexualized content.
A spokesperson for the California Department of Justice said xAI, the company behind Grok, “can and should put better safeguards in place to protect children and women from the harms of sexually explicit materials being generated without their consent.” While welcoming the changes, officials confirmed the investigation is ongoing to determine whether any laws were violated.
The backlash has not been limited to the US. Indonesia and Malaysia became the first countries to suspend Grok entirely, while lawmakers in the UK openly discussed blocking the tool. Britain’s communications regulator Ofcom has launched its own investigation into the chatbot’s image-generation capabilities.
Safety concerns expose divide in AI industry
UK technology secretary Liz Kendall said she welcomed X’s action but stressed that regulators would still establish the full facts through Ofcom’s inquiry. Critics have argued that earlier restrictions, which limited image generation to paid users, merely shifted the risk rather than removing it.
A spokesperson for Prime Minister Keir Starmer previously warned that turning controversial AI features into premium products did little to address the underlying harm.
Elon Musk, who owns xAI and X, pushed back against regulatory criticism, suggesting governments were seeking “any excuse for censorship.” He also encouraged users to attempt to bypass Grok’s safeguards shortly before the new restrictions were announced.
Calls for stronger AI safeguards
AI and cybersecurity experts say the controversy highlights a growing split in how companies approach AI governance. Dipal Dutta, CEO of UK-based IT firm Redoq, said safeguards should be embedded “at the point of origin,” including blocklists, cleaner training data, and secondary AI systems to detect and stop harmful content before it is generated.
Sarah Armstrong-Smith, a member of the UK government’s cyber advisory board, said the situation reflects a broader divide between firms prioritizing responsible AI and those framing resistance to regulation as a free speech issue.
She added that the Grok backlash could mark “the end of unchecked AI experimentation,” as governments become less willing to tolerate the social risks posed by powerful generative models.
Sources: Business Insider, X Safety Blog, California Department of Justice, Ofcom
