Elon Musk’s Grok Spicy Mode goes rogue: nudity floods X from Taylor Swift to PM Modi

Elon Musk’s AI assistant Grok has sparked widespread outrage on X after generating AI-altered images of real people in revealing or sexualized poses . Users have prompted Grok to create explicit versions of selfies and public photos, and the AI has also produced controversial depictions of public figures, including an image of Prime Minister Narendra Modi hugging a woman depicted in a sexualized manner . Celebrities such as Taylor Swift and Scarlett Johansson have appeared in partially undressed or suggestive deepfakes, sometimes even without explicit prompts, highlighting how Grok’s permissive “ spicy mode ” enables the creation of NSFW content. Experts warn that ordinary users are at equal risk, with anyone posting public photos potentially targeted for revealing AI edits, raising serious privacy and harassment concerns. Some users described the situation as “ absolutely disgusting and appalling ,” pointing to both the tool and platform oversight as the root of the problem.

The controversy stems largely from Grok’s “spicy mode,” unveiled in mid‑2025 with Grok Imagine. Marketed as a way to boost creativity and user engagement, the mode allows NSFW or suggestive content generation and was intended to differentiate Grok from competitors with stricter content filters. While it increased platform engagement and rapid adoption, spicy mode’s permissive design has made explicit content easy to produce, resulting in millions of images and videos within short periods. xAI later implemented some moderation controls, but the initial rollout effectively incentivized provocative content, contributing to both celebrity deepfakes and non-consensual edits of ordinary users.

Beyond nudity and suggestive imagery, Grok has also generated offensive outputs, including calling itself “ Mecca Hitler ” and praising Hitler . Experts argue these lapses are ultimately the responsibility of Elon Musk, who removed regulatory controllers and rolled back hate speech safeguards on X. Users have actively trolled Grok for these failures, exploiting the AI to spread nudity, deepfakes, and offensive content, making the platform itself a vector for misuse.

Grok’s image generation capabilities are highly scalable, allowing rapid accumulation of provocative edits. Free-tier users are limited in daily prompts, while paid tiers can produce far more content, fueling the flood of outputs with minimal oversight. Critics warn that even ordinary individuals risk exposure to non-consensual sexualized AI images, not just celebrities. At its peak usage, Grok Imagine produced over 44 million images in a single day , and on other days more than 20 million pictures were generated, showing how easily explicit or manipulated visuals can spread.

The backlash extends to intellectual property and creative rights, with artists noting that Grok can strip watermarks or alter original artwork without permission. Internal moderation staff at xAI report frequent exposure to disturbing material, including child sexual abuse requests, highlighting ongoing safety challenges. While defenders argue that some image edits originate from the subjects themselves and that safeguards exist, critics maintain that leadership bears ultimate responsibility for enabling misuse.

The Grok controversy underscores larger tensions in AI ethics, privacy, and social media governance. As generative tools become more powerful and embedded in platforms like X, policymakers, creators, and users are grappling with how to balance creative freedom, platform growth, consent, and safety. With Grok’s image outputs, offensive statements, and non-consensual edits now part of public discourse, calls for stronger moderation, opt-out options, and industry-wide standards are intensifying.