According to multiple former employees, Elon Musk's AI company xAI is doubling down on Grok's video and image generation capabilities — and the results are, well, extremely NSFW. Two recently departed staffers estimate that more than half of Grok's traffic comes from adult images, videos, and roleplay conversations.
Even Grok's code generation model is being heavily repurposed to churn out adult imagery, mainly because the code model is cheaper to run. The NSFW subscriptions are bringing in serious cash: the premium tier, Grok Imagine, costs a whopping $300 per month. SpaceX's IPO filing reveals that its AI division posted $3.2 billion in revenue in 2025, and by Q1 2026, it was generating 10 billion images and 2 billion videos per month.
But the lax content policies are drawing regulatory heat and internal strife. The EU, France, and the UK have launched investigations into Grok generating child sexual abuse material. Inside xAI, the ability to create realistic nude images has researchers worried, and some tech staff have quit after being reassigned to develop a sexy avatar named "Ani." SpaceX's IPO filing warns that Grok's "frank and uninhibited" mode could lead to reputational damage and explicit imagery.
Longer term, Grok's video annotation data is being fed into a "world model" that predicts real-world environments — potentially destined for Tesla's self-driving and the Optimus robot. Meanwhile, competitors are clamping down hard on adult content. Anthropic explicitly bans Claude from generating sexually explicit material; Google and Midjourney block adult content in their video and image models; and OpenAI has shelved its own erotic content plans.