Elon Musk’s AI chatbot Grok is at the center of one of the biggest AI safety controversies of the year. After weeks of mounting pressure, X (formerly Twitter) has announced sweeping restrictions on Grok’s image-generation tool, following a global backlash over its creation of sexualized deepfakes of real people — including women and children.
The move marks a dramatic reversal for xAI, Musk’s artificial intelligence company, which had rolled out Grok’s image editing features with minimal safeguards. What followed was described by California’s attorney general as a “shocking” avalanche of non-consensual, sexually explicit material.
What happened: “Spicy Mode” and simple text prompts
The controversy centers on Grok’s image editing capabilities, including a so-called “Spicy Mode” that allowed users to manipulate photos with simple text prompts such as “put her in a bikini” or “remove her clothes.”
According to reports, users weaponized the tool to generate vast numbers of sexualized images of public figures and private citizens. One analysis estimated around 1.8 million non-consensual sexualized deepfakes flooded social media within days of the feature’s broad rollout.
The victims include high-profile figures. Labour MP Jess Asato revealed that users ordered the bot to create images of her in bikinis, and she has since launched what is believed to be the first UK legal claim against xAI over non-consensual AI-manipulated imagery. In another disturbing case, a parent reported that Grok had inappropriately altered a photo of her child.
Regulators around the world respond
The backlash quickly went global, with governments and regulators on multiple continents taking action:
- United States: California Attorney General Rob Bonta launched an investigation into xAI, stating his office has “zero tolerance for the AI-based creation and dissemination of nonconsensual intimate images or of child sexual abuse material.”
- European Union: EU digital affairs spokesperson Thomas Regnier confirmed the bloc was seriously examining complaints, noting that some outputs involved sexualized images of children — calling the feature illegal and unacceptable in Europe.
- United Kingdom: Media regulator Ofcom urgently contacted X and xAI and is investigating whether the platform breached the Online Safety Act. Prime Minister Sir Keir Starmer publicly condemned the spread of sexualized AI-generated images.
- India: Authorities ordered X to remove sexualized content and submit an action report within 72 hours.
- Malaysia and Indonesia: Both countries took legal action, with access to Grok blocked in some jurisdictions.
- France and Brazil: Issued warnings and called for investigations into Grok’s misuse.
X’s response: geoblocking and paid-only access
Facing this unprecedented pressure, X announced a series of restrictions. The platform said it will “geoblock the ability” of all Grok and X users to create images of real people in “bikinis, underwear, and similar attire” in jurisdictions where such actions are illegal.
“We have implemented technological measures to prevent the Grok account from allowing the editing of images of real people in revealing clothing,” X’s safety team said, adding that the restriction applies to all users, including paid subscribers.
In what the company called “an extra layer of protection,” image creation and photo editing via Grok is now limited to paid subscribers only — a move X says will ensure that anyone who abuses the tool “can be held accountable.”
The European Commission said it had taken note of the additional measures and would carefully assess whether they effectively protect EU citizens. Ofcom welcomed the development but stressed its investigation continues “round the clock.”
Why this matters for the future of AI
The Grok controversy has become a defining case study in the debate over AI safety. Critics argue that xAI bypassed the rigorous safety testing and red-teaming protocols used by other leading AI labs, deploying an image pipeline that was easily abused from day one. For context on how fast AI is moving, see our coverage of Google’s Gemini AI.
Investigations also raised questions about Grok’s underlying safety instructions. Reports indicated that Grok’s system prompts at one point stated there were “no restrictions” on certain types of adult content — a stark contrast to the guardrails maintained by competitors.
For lawmakers, the episode provides powerful ammunition to fast-track stricter regulations on AI-generated content. The incident is already being cited in discussions about liability frameworks for AI outputs — potentially rewriting the rules for how AI companies deploy powerful generative tools.
What’s next for Grok and xAI
While X’s new restrictions represent a significant climbdown, serious questions remain. Regulators in the UK, EU, and California are continuing their investigations, and legal claims from victims are working their way through courts. Whether geoblocking and paywalls are enough to prevent further abuse — or whether tougher legislation will follow — remains to be seen.
One thing is clear: the era of deploying powerful AI image tools without meaningful safeguards is over. The Grok backlash will be remembered as the moment the world demanded accountability from AI companies — and started to get it.
What do you think — are X’s new restrictions enough, or should AI image tools face stricter laws worldwide? Share your thoughts in the comments.