\u201cSince discovering Grok AI, regular porn doesn\u2019t do it for me anymore, it just sounds absurd now,\u201d one enthusiast for the Elon Musk-owned AI chatbot wrote on Reddit. Another agreed: \u201cIf I want a really specific person, yes.\u201d<\/p>\n
If those who have been horrified by the distribution of sexualised imagery on Grok<\/a> hoped that last week\u2019s belated safeguards<\/a> could put the genie back in the bottle, there are many such posts on Reddit and elsewhere that tell a different story.<\/p>\n And while Grok has undoubtedly transformed public understanding of the power of artificial intelligence, it has also pointed to a much wider problem: the growing availability of tools, and means of distribution, that present worldwide regulators with what many view as an impossible task. Even as the UK announces that creating nonconsensual sexual and intimate images will soon be a criminal offence, experts say that the use of AI to harm women has only just begun.<\/p>\n Other AI tools have much stricter safeguards in place. Asked to strip a photograph of a woman into a bikini, the large language model (LLM) Claude says: \u201cI can\u2019t do that. I\u2019m not able to edit images to change clothing or create manipulated photos of people.\u201d ChatGPT<\/a> and Google\u2019s AI tool Gemini will create bikini images, but nothing more explicit.<\/p>\n However, there are far fewer limits elsewhere. Users of the Grok forum on Reddit<\/a> have been sharing tips on how to generate the most hardcore pornographic images possible using pictures of real women. On one thread, users were complaining that Grok would allow them to make images of women topless \u201cafter a struggle\u201d, but refused to generate genitals. Others have noticed that asking for \u201cartistic nudity\u201d gets around safeguards around stripping women completely naked.<\/p>\n Grok has also been used to generate deepfake images of Elon Musk in a bikini. Photograph: Leon Neal\/Getty Images<\/p>\n Beyond LLMs and major platforms is a whole ecosystem of websites, forums and apps devoted to nudification and the humiliation of women. These communities are increasingly finding pipelines to the mainstream, said Anne Craanen, a researcher at the Institute for Strategic Dialogue (ISD) working on tech-facilitated, gender-based violence.<\/p>\n Communities on Reddit and Telegram discuss how to bypass guardrails to make LLMs produce pornography, a process known as \u201cjailbreaking.\u201d Threads on X amplify information about nudification apps<\/a>, which produce AI-generated images of women with their clothing removed, and how to use them.<\/p>\n Craanen said the route for misogynistic content to reach the wider internet has grown broader, adding: \u201cThere is a very fruitful ground there for misogyny to thrive.\u201d<\/p>\n Research from the ISD last summer found<\/a> dozens of nudification apps and websites, which collectively received nearly 21 million visitors in May 2025. There were 290,000 mentions of these tools on X in June and July last year. Research by the American Sunlight Project in September found that there were thousands of ads<\/a> for such apps on Meta, despite the platform\u2019s efforts to crack down<\/a> on them.<\/p>\n \u201cThere are hundreds of apps hosted on mainstream app stores like Apple and Google that make this possible,\u201d said Nina Jankowicz, a disinformation expert who co-founded the American Sunlight Project. \u201cMuch of the infrastructure of deepfake sexual abuse is supported by companies that we all use on a daily basis.\u201d<\/p>\n