NSFW Content Accounts for 'Well Over Half' of Grok Traffic, Per Former xAI Employees

What the Numbers Say

Adult content is Grok's biggest product, whether xAI wants to advertise that or not.

According to The Information, which cited two former xAI employees, NSFW uses including pornography generation, adult role-play chats, and erotica requests account for "well over half" of all traffic to the Grok platform. That's the core business.

Users have reportedly found a workaround to reduce costs: routing explicit image and text requests through xAI's coding-focused models, which are priced lower than the general-purpose versions. An internal analysis found that a "significant proportion" of requests to that coding model were for porn or nude images, according to The Information's sources.

What xAI Told Investors — and What It Left Out

xAI filed IPO paperwork that acknowledged Grok's more "irreverent" features as a potential risk to the business. The company set aside $530 million to cover potential legal costs related to that exposure.

What the paperwork did not do, according to The Information's reporting, is specifically identify NSFW activity as a meaningful revenue driver. Even though, if the traffic figures hold, it is almost certainly one of the largest contributors to the company's usage metrics and, by extension, its revenue.

IPO disclosures are supposed to give prospective investors a clear picture of how a business actually makes money and what risks that revenue carries. Framing adult content as a general liability without disclosing it as a primary revenue source raises legitimate questions about how complete that picture was.

The CSAM Problem Was Not Simple to Fix

xAI's engineers faced a specific technical and ethical problem: enabling adult content while building hard walls against child sexual abuse material. The Information's sources said there were no "quick fixes" for that dilemma.

Content moderation at the edge of explicit-but-legal versus illegal material is one of the hardest problems in platform technology. The fact that a well-funded AI company found it genuinely difficult speaks to the scope of the engineering and moral stakes involved.

Some employees were reportedly "embarrassed and disturbed" by the work. The Engadget writeup of The Information's report notes that a group of engineers were assigned to build "Ani," an NSFW anime-style avatar companion, and that the internal reaction was far from enthusiastic.

The Sexualized Real-People Problem

Grok generated sexualized images of real people, including minors. That sparked multiple lawsuits and investigations, according to Engadget. X subsequently limited Grok's ability to produce racy edits of real people's images directly on the platform. But paid xAI subscribers can still generate such material, per tests Engadget says were conducted.

Limiting the free on-platform tool while keeping the capability alive for paying customers is a business decision dressed up as a safety measure.

The Government Contract Problem

xAI has pursued and secured contracts with U.S. government agencies and military institutions, according to Engadget. The specifics of those contracts — which agencies, what the dollar values are, and what data or systems are involved — are not detailed in the available sourcing.

The tension is obvious. Federal agencies and the military are not typically in the market for AI systems whose primary traffic load is pornographic. Whether the agencies contracting with xAI are using entirely separate, ringfenced model deployments, or whether there's any operational overlap with the consumer-facing Grok infrastructure, is not publicly established.

The Strongest Defense of xAI's Approach

The legitimate argument in xAI's favor is this: adults seeking legal adult content are going to find it somewhere, and a platform that serves that demand while attempting to build meaningful CSAM guardrails is arguably preferable to unmoderated alternatives. Fewer restrictions on legal content between consenting adults is a coherent free-speech and personal-liberty position. xAI leaning into that market is a business decision, not inherently a moral failure.

The problem is that the sexualized-images-of-real-people episode, particularly involving minors, goes well beyond consenting adults accessing legal content. That's where the defense collapses. And the ongoing availability of that capability for paid subscribers, per Engadget's testing, suggests the fix applied was partial at best.

The Unresolved Question

The lawsuits and investigations stemming from Grok's deepfake and CSAM incidents are ongoing, according to Engadget. What remains genuinely unknown is whether any of those legal actions will affect xAI's existing government contracts or its ability to win new ones. According to Engadget, it is not clear whether those incidents have had any impact on xAI's relationship with the federal government. Federal procurement officers have access to the same public reporting everyone else does. Whether that reporting changes any contract calculus is the question xAI's leadership will have to answer, probably not on their own timeline.