xAI's Grok is hosting nonconsensual sexualized deepfakes of real women on its public website — and has not removed them.
An investigation found dozens of AI-generated "nudified" images and videos of real people hosted directly on Grok's platform. The subjects include celebrities and at least one prominent US politician. The framing of the investigation matters: this is not a first-detection story. Grok is still doing this — which implies either a content moderation process that doesn't work or one that hasn't been seriously tried.
The presence of a politician sharpens the legal exposure. Federal lawmakers have spent the past two years building statutory frameworks around nonconsensual intimate imagery, AI-generated or otherwise, and the industry has broadly moved to get ahead of that. Most major AI labs now treat this category of content as a hard line. xAI's apparent inaction isn't just a reputational problem — it's a meaningful divergence from where the rest of the sector has landed.
xAI's owner has made minimizing content moderation a brand value. That argument is easier to sustain when the speech in question isn't nonconsensual pornography of named individuals.