This is a real and serious problem area, so it gets an honest “mixed.” There have been genuine, well-documented incidents of xAI’s Grok Imagine producing sexualised or nonconsensual images, including of public figures, and that is indefensible — image generators across the industry have struggled with exactly this. What the strongest version of the claim gets wrong is intent and design: Grok’s acceptable-use policy explicitly prohibits sexual content depicting real, identifiable people without consent, the system blocks many such prompts, and xAI has rolled out remediations as failures surface. So the truthful framing is neither “it’s fine” nor “it’s purpose-built for abuse”: it’s that a powerful generative tool has produced harmful outputs that violate its own rules, and the legitimate criticism is about how robustly xAI enforces those rules — a safety-engineering failure to fix, not a stated design goal.