The initial excitement around Grok Spicy Mode has been quickly overshadowed by major controversy. As reported by The Verge and TechCrunch, the tool was used to create non-consensual deepfake videos of celebrities like Taylor Swift, often without an explicit NSFW prompt. Simply asking for a celebratory video could result in a partially nude animation.
This has triggered a significant public backlash and, predictably, a tightening of xAI's moderation. Our tests confirm this; the chances of successfully generating a truly NSFW video are now extremely low. Given the legal risks and public pressure from laws like the Take It Down Act, it's almost certain that Grok's restrictions will only get stricter. For creators, this makes Grok an unreliable and frustrating platform for NSFW content.
So, how are creators navigating this new, stricter environment? It has become a game of cat-and-mouse, requiring creative workarounds with no guarantee of success. Some users have found limited success by using indirect, "poetic" prompts to describe scenes, avoiding direct trigger words. Others use a "bait and switch" method: first generating a safe but suggestive image, then applying a more direct prompt in the video animation stage. While these tricks can sometimes bypass the filters, they are inconsistent and time-consuming, turning the creative process into a frustrating puzzle.