From the second Elon Musk’s artificial intelligence firm, xAI, started rolling out its Grok chatbot to paid X subscribers in 2023, it pitched the device because the dangerous boy of enormous language fashions. Grok would supposedly be licensed to say and do issues that its politically right opponents—primarily ChatGPT, produced by Musk’s previous nemeses at OpenAI—wouldn’t.
In an announcement on X, the corporate touted Grok’s “rebellious streak” and teased its willingness to reply “spicy” questions with “a little bit of wit.” Though xAI warned that Grok was “a really early beta product,” it assured customers that with their assist, Grok would “enhance quickly with every passing week.”
On the time, xAI didn’t promote that Grok would in the future ship nonconsensual pornography on an on-demand foundation. However over the previous few weeks, that’s precisely what has occurred, as X subscribers inundated the platform with requests to change actual pictures of girls by eradicating their clothes, altering their our bodies, spreading their legs, and so forth. X customers don’t have to be premium subscribers to avail themselves of those companies, that are accessible each on X and on Grok’s stand-alone app.
Some pictures generated with Grok’s help depict topless or in any other case suggestive pictures of women between ages 11 and 13, according to a U.Ok.-based baby security watchdog. One analysis of 20,000 pictures generated by Grok between December 25 and January 1 discovered that the chatbot had complied with person requests to depict kids with sexual fluids on their our bodies. On New 12 months’s Eve, an AI agency that provides picture alteration detection companies estimated that Grok was churning out sexualized pictures at a fee of about one per minute.
“I’ve been sexually assaulted previously, and it virtually felt like a digital model of that,” one lady told The Reduce after Grok customers remodeled an image of her posing subsequent to a Christmas tree whereas carrying exercise gear into an image of her carrying a thong bikini. “It’s unfathomable to me that persons are allowed to do that to girls.”
On Thursday, journalist and Bellingcat founder Eliot Higgins reported seeing Grok-generated pictures of Renee Nicole Good, the unarmed 37-year-old lady shot and killed by ICE brokers in Minneapolis, altered to depict her lifeless physique in a bikini. As Higgins put it: “Digital corpse desecration now out there to the general public.”
For all of the potential use instances of AI chatbots that AI firms have touted lately, bespoke pornography was all the time the howlingly apparent one. (You don’t have to be a behavioral scientist to know what sure demographics immediately think to do when offered with a device marketed as able to magically producing photorealistic pictures of something one’s thoughts can dream up.) With various levels of success, platforms like ChatGPT and Google’s Gemini have at the least tried to get forward of this eventuality, constructing “guardrails” that attempt to restrict customers’ capability to customise NSFW pictures to go well with their tastes.
A serious distinction between these firms and xAI, after all, is that xAI is helmed by Musk, whose ideological dedication to eradicating wokeness and censorship extends to providing amused, winking defenses of nonconsensual grownup content material revealed by his firm’s flagship product.
On his X account, Musk has been firing off prompts treating what could be an existential disaster for another firm as a enjoyable and humorous meme. The truth that one of many victims was Ashley St. Clair, the mom of Musk’s 1-year-old son, didn’t dissuade Musk from declaring himself unable to cease laughing at an AI picture of a bikini-clad toaster.
Each X and Musk have since issued statements reminding customers that the X phrases of service bar the creation of kid sexual abuse materials (CSAM) and pornography. X has additionally stated that it removes CSAM and different unlawful content material, and completely suspends accounts that create it. On the identical time, it’s kind of difficult for the corporate to place itself as taking an issue significantly when its proprietor, who can be the most-followed particular person on the platform, was logging on and treating your entire factor as one large joke.
Usually, the existence of a web-based device able to producing one-click CSAM would immediate widespread outrage and speedy responses from regulation enforcement. Regulators in international locations in Europe, Asia, and South America have promised to research, and this week the European Fee extended an order that requires X to retain all Grok-related paperwork and information whereas officers take a better look.
There are present authorized mechanisms within the U.S. for addressing the vile issues Grok is doing, too. Lower than a 12 months in the past, for instance, Trump signed into regulation the TAKE IT DOWN Act, a bipartisan invoice that requires web sites to take away “nonconsensual intimate imagery” inside 48 hours upon the sufferer’s request. And though a provision of federal regulation referred to as Part 230 typically protects web sites and social media platforms from legal responsibility for content material revealed by their customers, right here, Grok itself is doing the “publishing” by producing the pictures.
Sen. Ron Wyden (D-OR), who helped write Part 230 three a long time in the past, weighed in on Bluesky, arguing that “AI chatbot outputs will not be protected” beneath Part 230, and that “it’s not a detailed name.” Together with two different Democratic senators, Wyden has additionally asked Apple and Google to take away Grox and X from their app shops for violations of the businesses’ phrases of service. This might be a major step past what seems to be the one motion taken by Apple so far: raising its age ranking of the Grok app from 12+ to 13+.
All that stated, Musk, who spent 4 whirlwind months hacking away on the administrative state as head of the Division of Authorities Effectivity, has loads of sensible causes to not be fearful. Because of the political and monetary assist that Musk and his Silicon Valley friends present to the Republican Social gathering, the second Trump administration has been enthusiastic about integrating AI merchandise—each from xAI and from different firms—into the workings of the federal authorities.
The truth that Trump instantly designated David Sacks, a tech investor with vital AI and crypto pursuits (in addition to shut private {and professional} ties to Musk), as his AI and crypto czar is a reasonably good indication that significant regulation will not be coming anytime quickly.
Since 2019, states with each Democratic- and Republican-controlled legislatures have responded to the absence of federal motion by passing greater than 140 state legal guidelines regulating AI, in line with a Brennan Heart analysis. However in December 2025, the White Home made what is maybe its most promising gesture but to the AI trade: an executive order reiterating Trump’s dedication to a constructing “minimally burdensome nationwide coverage framework for AI” that may “maintain and improve the US’ international AI dominance.”
Amongst different issues, the order directs govt businesses to determine state AI rules that the administration deems inconsistent with its agenda, and encourages Legal professional Normal Pam Bondi to type an “AI Litigation Process Drive” to problem the offending legal guidelines in courtroom.
Like most Trump govt orders, this one is not going to have the quick influence that some breathless headlines recommend; because the order itself acknowledges, Congress would want to behave to ensure that the substantive provisions to take impact. However for Musk, the message the White Home is sending about its priorities is what actually issues: Proper now, the Trump administration is simply too preoccupied with beginning unlawful wars and executing unarmed protesters within the streets to fret about a number of risqué pictures showing on its social media platform of selection.
When Musk left Washington final 12 months, he didn’t achieve this quietly, lashing out at Trump for being insufficiently deferential to his preferences and insufficiently grateful for his assist. However eight months later, the truth that the official response to Grok’s pornography and CSAM options is successfully a disinterested shrug demonstrates that the quarter-million {dollars} Musk donated to Trump and different Republicans in 2024 was a sound funding in his firm’s future.
By January 3, whereas Grok was nonetheless spitting out these pictures upon request, Musk and Trump had reconciled sufficient to have dinner collectively at Mar-a-Lago. Afterward, Trump called Musk “nice” and “ man,” and Musk predicted that 2026 could be “superb.”
Legal guidelines are solely as sturdy because the willingness of the powers that be to implement them. Whenever you personal the individuals who regulate you, there isn’t a scandal too disgusting so that you can snigger off.
