NYT and CCDH analysis: Grok created and shared 1.8M+ sexualized images of women between December 31 and January 8, after Elon Musk promoted the feature on X
Over nine days, Elon Musk's Grok chatbot generated and posted 4.4 million images, of which at least 41 percent were sexualized images of women.
That progression matters because promotion and distribution were intertwined on the same service: image-generation capability could create harmful material and publish it into X’s social feed in one workflow.
First-order effects
The findings put xAI and X under sharper scrutiny over whether their image-generation controls and removal processes can prevent sexualized depictions of women at the volume documented.
Women whose likenesses are altered or sexualized face a larger exposure surface when generation and sharing are combined inside a major social platform.
Second-order effects
The reported scale raises the operational bar for competing AI products and social platforms: prompt controls alone are less credible if generated images can be rapidly distributed or regenerated.
Advertisers, users, and trust-and-safety teams gain a concrete measure against which to assess claims that safeguards have addressed the earlier failures.
Third-order effects
If integrated AI creation and social distribution continue to expand together, safety will increasingly be judged as a product-and-distribution governance problem rather than solely a model-output problem.
The episode fits a broader shift toward accountability for the downstream harms of generative features, especially where consent, identity, and rapid public sharing intersect.
The trend:Generative AI platforms are being evaluated not just on what models can produce, but on whether their distribution systems can contain repeatable, nonconsensual misuse at scale.
New research from the Center for Countering Digital Hate claims to put into context the number of sexualized images Grok created: — “Grok produced an estimated 3 million sexualized images including 23,000 of children”
🚨 NEW: Around 3 MILLION sexualized images over just 11 days? That's what our findings estimate X's Grok AI generated. — CCDH researchers saw an unprecedented flood of photorealistic, sexualized images, including children, after X made the existing feature even easier to access…
Grok AI generated an estimated 190 sexualized images every single minute. — Elon Musk's AI tool is utterly failing to implement strong safeguards, seriously risking the safety of women and kids. — And the news for children gets even worse... [image]
So we have figures now. — Grok created and shared 1.8m sexualised images of women between December 31 and January 8, after Elon Musk promoted the feature on X. — Overall, the Grok chatbot generated and posted 4.4m images — 41% were sexualised images of women — www.nytimes.c…
And no questions about Musk's Grok AI and why it's been creating nonconsensual sexualized image women and kids. I guess a private equity CEO isn't the best equipped to ask seemingly tough questions. — For more: www.nytimes.com/2026/01/22/t...
Outside of the antebellum South, I'm hard pressed to think of an American elite that would have embraced and celebrated this kind of fully-automated depravity in the way today's billionaires and their servants have.
The @nytimes.com and @counterhate.com dug into the massive output of sexualized AI images created by X's AI chatbot, Grok. — Based on their data analysis, they estimated that, in just a few weeks, Grok had generated *23,000* images of sexualized children. — www.nytimes.com/20…
Read the full report here 👇 — counterhate.com/research/gro... And read my interview with CCDH's CEO Imran Ahmed, who has been threatened with deportation by the US administration 👇 — www.politicshome.com/news/article...