Grok Imagine's NSFW 'Spicy Mode' Sparks AI Content Controversy

xAI, Elon Musk’s artificial intelligence company, has officially launched Grok Imagine, a new feature designed to generate images and short videos from text prompts. The tool is now accessible to SuperGrok and Premium+ subscribers on X (formerly Twitter) via iOS. In line with Musk’s stated “no-filter” approach to content, Grok Imagine permits the creation of content that includes semi-nude imagery.

Grok Imagine is capable of producing 15-second videos, complete with native audio, from either text or existing image prompts. A notable inclusion is its “spicy mode,” a setting that broadens the scope for adult-themed content creation. While the feature strictly prohibits fully explicit or pornographic visuals, users have successfully generated AI-created images depicting partial female nudity. However, the system’s moderation filters are designed to blur or block content deemed excessively graphic.

This permissive stance on content aligns with xAI’s recent release of a highly sexualized anime-style AI companion. Nevertheless, the introduction of Grok Imagine’s “spicy mode” could ignite controversy, echoing past criticisms faced by the Grok chatbot, which previously generated antisemitic and misogynistic content.

Public awareness of the “spicy mode” first emerged following a now-deleted X post by xAI employee Mati Roy, who indicated the feature’s capability for “nudity.” Since then, users have actively tested the tool’s boundaries.

Despite its openness to certain types of mature content, Grok Imagine does impose specific restrictions. For instance, attempts to generate bizarre or inappropriate images involving public figures, such as a pregnant Donald Trump, are blocked, with the tool instead offering more innocuous alternatives. Similarly, generating images of public figures like Taylor Swift is possible, but within defined limits.

From a technical standpoint, Grok Imagine demonstrates impressive capabilities. It rapidly generates images from prompts, automatically creates new visuals as users scroll, and can transform still images into short, stylized videos. While human figures in the generated content sometimes appear somewhat artificial, characterized by a waxy skin texture and cartoon-like features, the tool shows significant potential. Elon Musk has stated that the model is expected to “get better every day.”

As Grok Imagine continues to push creative and ethical boundaries, the central question remains how far the technology will evolve and what, if any, content guidelines it will ultimately adhere to or transgress.