A correction to my earlier post. I was right that the viral @Grok episode on X chiefly involved bikini, underwear and sexualised-pose edits, and that xAI subsequently closed the easy public route. I was too categorical in saying Grok had never generated pornography. AI Forensics documented full nudity and sexual acts through Grok.com, while the Jane Doe 4 lawsuit alleges about 7,000 explicit outputs. That allegation is not yet a court finding, but it cannot honestly be dismissed.
That still does not establish the thread title or the claims about Musk’s motives. A serious product-safety failure is not proof that Musk is “the biggest exploiter of women online”, deliberately built a porn bot or wants women abused. In the Wyoming allegation, the perpetrator was the stepfather using the tool. xAI can fairly be criticised for making image editing too easy and for inadequate initial safeguards. Musk’s personal malice and a pornography-based business plan are assumptions, not evidence.
The chronology also matters. xAI blocked the viral @Grok route in January, and X subsequently told Parliament that its nudification functionality had been disabled worldwide. xAI’s current rules expressly prohibit nudifying real people, pornographic depictions of their likenesses and sexualising children.
A later security report showed that Grok’s protections could be jailbroken, but the same jailbreak affected ChatGPT and Axios could not reproduce the Grok result in its own repeated testing. A jailbreak is a safety problem. It is not the same thing as a publicly supported feature.
Most importantly, Grok is not the whole problem. ISD found 181 dedicated nudification sites attracting more than 40 million visitors each month, alongside downloadable and locally operated models that no company can centrally disable. These decentralised tools are the harder long-term enforcement problem.
None of this minimises non-consensual sexual images. It identifies the danger accurately. Anger at Musk for unsupported reasons hands people an easy factual rebuttal and distracts from the dedicated, decentralised systems that are much harder to police.
I also think Grok deserves recognition for being unusually willing, among mainstream chatbots, to discuss biological sex and sex-based rights without automatically adopting gender-identity framing. That is my assessment, not a scientific league table. It does not excuse xAI’s early safety failure, and that safety failure does not make every accusation about Musk true.