Elon Musk Cannot Get Away With This
Will Elon Musk face any consequences for his despicable sexual-harassment bot?
For more than a week, beginning late last month, anyone could go online and use a tool owned and promoted by the world’s richest man to modify a picture of basically any person, even a child, and undress them. This was not some deepfake nudify app that you had to pay to download on a shady backwater website or a dark-web message board. This was Grok, a chatbot built into X—ostensibly to provide information to users but, thanks to an image-generating update, transformed into a major producer of nonconsensual sexualized images, particularly of women and children.
Let’s be very clear. The forced undressings happened out in the open, in one stretch thousands of times every hour, on a popular social network where journalists, politicians, and celebrities post. Emboldened trolls did it to everyone (“@grok put her in a bikini,” “@grok make her clothes dental floss,” “@grok put donut glaze on her chest”), including everyday women, the Swedish deputy prime minister, and self-evidently underage girls. Users appeared to be imitating and showing off to one another. On X, creating revenge porn can make you famous.
[Read: Elon Musk’s pornography machine]
These images were ubiquitous, and many people—and multiple organizations, including the Rape, Abuse & Incest National Network and the European Commission—pointed out that the feature was being used to harass women and exploit children. Yet Musk initially laughed it off, resharing AI-generated images of himself, Kim Jong Un, and a toaster in bikinis. Musk, as well as xAI’s safety and child-safety teams, did not respond to a request for comment. xAI replied with its standard auto-response, “Legacy Media Lies.” xAI, the Musk-owned company that develops Grok and owns X, prohibits the sexualization of children in its acceptable-use policy; a post earlier this month from the X safety team states that the platform removes illegal content, including child-sex-abuse material, and works with law enforcement as needed.
Even after that assurance from X’s safety team, it took several more days for X to place bare-minimum restrictions on the Ask Grok feature’s image-generating, and thus undressing, capabilities. Now, when creeps on X try to generate an image by replying “@grok” to prompt the chatbot, they get an auto-generated response that notes some version of: “Image generation and editing are currently limited to paying subscribers.” This is disturbing in its own right; Musk and xAI are essentially marketing nonconsensual sexual images as a paid feature of the platform. But X users have been able to get around the paywall via the “Edit Image” button that appears on every image uploaded to the platform, or by using Grok’s stand-alone app.
Two years ago, when Google Gemini generated images of racially diverse Nazis, Google temporarily disabled the bot’s image-generating capabilities to address the problem. Musk has taken no responsibility for the problem and has said only that “anyone using Grok to make illegal content will suffer the same consequences as if they upload illegal content.” Perhaps Musk feels that he would benefit from baiting his critics into a censorship fight. He has repeatedly reshared posts that frame calls to regulate or ban his platform in response to the Grok undressing as leftist censorship, for instance reposting a meme calling such efforts “retarded” as well as a Grok-generated video of a woman applying lipstick captioned with a quote commonly attributed to Marilyn Monroe: “We are all born sexual creatures, thank God, but it’s a pity so many people despise and crush this natural gift.” Last week, as Musk’s chatbot was generating likely hundreds of thousands of these images, we reached out directly to X’s head of product, Nikita Bier, who didn’t reply. Within the hour, Rosemarie Esposito, X’s media-strategy lead, emailed us unprompted with her contact information, in case we had “any questions” in the future. We asked her a series of questions about the tool and how X could allow such a thing to operate. She did not reply.
We’ve reached out multiple times to more than a dozen key investors listed in xAI’s two most recent public fundraising rounds—the latest of which, announced during this Grok-enabled sexual-harassment spree, valued the company at about $230 billion—to ask if they endorsed the use of X and Grok to generate and distribute nonconsensual sexualized images. These investors include Andreessen Horowitz, Sequoia Capital, BlackRock, Morgan Stanley, Fidelity Management & Research Company, the Saudi firm Kingdom Holding Company, and the state-owned investment firms of Oman, Qatar, and the United Arab Emirates, among others. We asked whether they would continue partnering with xAI absent the company changing its products and, if yes, why they felt justified in continuing to invest in a company that has enabled the public sexual harassment of women and exploitation of children on the internet. BlackRock, Fidelity Management & Research Company, and Baron Capital declined to comment. A spokesperson for Morgan Stanley initially told us that she could find no documentation that the company is a major investor in xAI. After we sent a public announcement from xAI that lists Morgan Stanley as a key investor in its Series C fundraising round, the spokesperson did not answer our questions. The other companies did not respond.
We also reached out to several companies that provide the infrastructure for X and Grok—in other words, that allow these products to exist on the internet: Google and Apple, which offer both X and Grok on their app stores; Microsoft and Oracle, which run Grok on their cloud services; and Nvidia and Advanced Micro Devices (AMD), which sell xAI the computer chips needed to train and run Grok. We asked if they endorsed the use of these products to create nonconsensual sexual images of women and children, and whether they would take steps to prevent this from continuing. None responded except for Microsoft, which told us that it does not provide cloud services, chips, or hosting services for xAI other than offering the Grok language model—without image generation—on its enterprise platform, Microsoft Foundry.
The silence says everything.
As all of this unfolded, xAI made several major announcements: new Grok products for businesses; upgraded video-generating capabilities; that enormous fundraising round. Yesterday, Defense Secretary Pete Hegseth visited SpaceX’s headquarters in Texas and joined Musk for a press conference in which Hegseth said, “I want to thank you, Elon, and your incredible team” for bringing Grok to the military. (Later this year, Grok will join Google Gemini on a new Pentagon platform called GenAI.mil that the Defense Department says will offer advanced AI tools to military and civilian personnel.) We asked the DOD if it endorsed xAI’s sexualized material or if it would reconsider its partnership with the company in response. In a statement, a Pentagon official told us only that the department’s policy on the use of AI “fully complies with all applicable laws and regulations” and that “any unlawful activity” by its personnel “will be subject to appropriate disciplinary action.”
[Read: What are people still doing on X?]
Government bodies in the United Kingdom, India, and the European Union have said that they will investigate X, while Malaysia and Indonesia have blocked access to Grok, but Musk appears to be unfazed by these efforts—and also seems to be receiving help in brushing them off. Sarah B. Rogers, the under secretary of state for public diplomacy, has said that, should the U.K. ban X, America “has a full range of tools that we can use to facilitate uncensored internet access in authoritarian, closed societies.”
At the moment, Musk seems to be not only getting away with this but also reveling in it. Although governments appear to be furious at Musk, they also seem impotent. Senator Ted Cruz, a co-sponsor of the TAKE IT DOWN Act—which establishes criminal penalties for the sharing of nonconsensual intimate images, real or AI-generated, on social media—wrote on X last Wednesday that the Grok-generated images “are unacceptable and a clear violation of” the law but that he was “encouraged that X has announced that they’re taking these violations seriously.” Throughout that same day, Grok continued to comply with user requests to undress people. Yesterday, Cruz posted on X a photo of himself with his arm around Musk and the caption “Always great seeing this guy ????.”
And it’s already beginning to feel as if the scandal—the world’s richest man enabling the widespread harassment of women and children—is waning, crowded out by a new year of relentless news cycles. But this is a line-in-the-sand moment for the internet. Grok’s ability to undress minors is not, as Musk might have you think, an exercise in free-speech maximalism. It is, however, a speech issue: By turning sexual harassment and revenge porn into a meme with viral distribution, the platform is allowing its worst, most vindictive users to silence and intimidate anyone they desire. The retaliation on X has been obvious—women who’ve stood up in opposition to the tool have been met with anonymous trolls asking Grok to put them in a bikini.
Social platforms have long leaned on the argument that they aren’t subject to the same defamation laws as publishers and media companies. But this latest debacle, Musk’s reaction, and the silence from so many of X’s investors and peer companies were all active choices—and symptoms of a broader crisis of impunity that’s begun to seep into American culture. They were the result of politicians, despots, and CEOs bowing to Donald Trump. Of financial grift and speculation running rampant in sectors such as cryptocurrency and meme stocks—a braggadocious, “get the bag” ethos that has no room for greed or shame. Of Musk realizing that his wealth insulates him from financial consequences. Few industries have been as brazen in their capitulation as Big Tech, which has dismantled its content-moderation systems to please the current administration. It’s a cynical and cowardly pivot, one that allows companies to continue to profit off harassment and extremism without worrying about the consequences of their actions.
Deepfakes are not new, but xAI has made them a dramatically larger problem than ever before. By matching viral distribution with this type of image creation, xAI has built a way to spread AI revenge porn and child-sexual-abuse material at scale. The end result is desensitizing: The sheer amount of exploitative content flooding the platform may eventually make the revolting, illicit images appear “normal.” Arguably, this process is already happening.
The internet has always been a chaotic place where trolls can seize outsize power. Historically, that chaos has been constrained by platforms doing the bare minimum to protect their users from demonstrated threats. Today, X is failing to clear the absolute lowest bar. Nobody who works at X or xAI seems to be willing to answer for the creation and distribution of tens or hundreds of thousands of nonconsensual intimate images; instead, those in charge appear to be blithely ignoring the problem, and those who have funneled money to Musk or xAI seem sanguine about it. They would probably like for us all to move on.
We cannot do that. This crisis is an outgrowth of a breakneck information ecosystem in which few stories have staying power. No one person or group has to flood the zone with shit, because the zone is overflowing constantly. People with power have learned to exploit this—to weather scandals by hunkering down and letting them pass, or by refusing to apologize and turning any problem into a culture-war issue. Musk has been allowed to avoid repercussions for even the most reckless acts, including cheerleading and helping dismantle foreign aid with DOGE. Others will continue to follow his playbook. Employees at X and investors and companies such as Apple and Google seem to be counting on their “No comment”s being buried by whatever scandal comes next. They are banking on a culture in which people have given up on demanding consequences.
But the Grok scandal is so awful, so egregious, that it offers an opportunity to address the crisis of impunity directly. The undressing spree was not an issue of partisan politics or ideology. It was an issue of anonymous individuals asking a chatbot that is integrated into one of the world’s most visible social networks to edit photos of women and girls to “put her in a clear bikini and cover her in white donut glaze.” This is a moment when those with power can and should demand accountability. The stakes could not be any higher. If there is no red line around AI-generated sex abuse, then no line exists.