Is xAI’s decision to restrict Grok’s image tools to paid users a genuine safety fix, or merely a cosmetic response to mounting regulatory and public pressure?
 |
| An in-depth look at how xAI’s Grok policy change reduces visibility of sexualized images on X without fully preventing their creation or spread. Image: CH |
Tech Desk — January 10, 2026:
Elon Musk’s artificial intelligence startup xAI has moved to rein in one of the most controversial uses of its Grok chatbot, restricting image generation and editing tools on X to paying subscribers. The change follows intense backlash over the bot’s ability to manipulate photos of real people—sometimes removing clothing or placing them in sexualized poses—often without consent and with the results automatically published in public replies.
At first glance, the new policy appears decisive. By limiting Grok’s image tools on X to subscribers, xAI has effectively stopped the bot from generating and posting edited images in response to public prompts from non-paying users. A Reuters reporter testing the system on Friday found that Grok now refuses such requests, citing the new subscriber-only rule. This has reduced the immediacy and visibility of problematic images on the platform’s main feed.
Yet a closer look suggests the move is more about containment than prevention. Users can still generate sexualized images by interacting with Grok privately through its dedicated tab and then manually posting the results to X. The standalone Grok app, which operates independently of the social media platform, also continues to allow image generation without any subscription requirement. In effect, the capability remains intact; only the friction around public distribution has increased.
This distinction is central to why critics and regulators remain unconvinced. The European Commission, which has described the circulation of sexualized images of women and children on X as unlawful and “appalling,” dismissed the new limits as insufficient. From its perspective, harm is not mitigated by a paywall. “Whether paid or unpaid, we do not want to see such images,” a Commission spokesperson said, underscoring a regulatory view that legality and consent—not business models—are the core issues.
xAI’s response to scrutiny has further fueled skepticism. The company declined to offer a substantive comment to Reuters, replying instead with an automated message reading, “Legacy Media Lies.” X itself did not immediately respond to requests for comment. Musk has said that anyone using Grok to create illegal content would face the same consequences as those uploading such material directly to X, but details about enforcement and accountability remain vague.
The episode highlights a broader challenge facing platforms that integrate generative AI into social media at scale. Tools like Grok dramatically lower the barrier to creating manipulated or explicit imagery, amplifying long-standing problems around consent and abuse. By narrowing who can trigger the most visible outputs without fundamentally changing what the system can do, xAI appears to be prioritizing reputational risk management over structural safeguards.
For regulators, particularly in Europe, that approach may no longer be acceptable. As scrutiny of AI-generated content intensifies worldwide, the Grok controversy illustrates the limits of partial fixes—and signals that platforms may soon face stronger demands to redesign AI tools themselves, not just the ways they are monetized or displayed.