Elon Musk's no-holds-barred AI chatbot Grok is again having one of its holds barred as regulators across the globe are balking at its widespread production of sexualized images, based on real images posted to X.
If you don't use Xitter anymore, first of all, who can blame you, but secondly you may not be aware that, as part of xAI now, the platform is a native home to Grok, xAI's chatbot, which users can simply send public @-messages to on X in order to ask it questions, or ask that it perform editing tasks with images. And who would've guessed that Grok's probably largely male userbase is asking it to turn many ordinary images of women sexual, saying things like, "Make the woman in in this picture turn around and bend over," and the like.
The issue was called out by a female gamer with over 6,000 followers a couple weeks back, as the New York Times reports, when this happened to her after she posted a photo of herself, and she asked, "Why is this allowed?"
Well, as you may recall, CEO Elon Musk pledged to create an un-PC, uncensored chatbot freed of the liberal guardrails imposed by the first chatbots out of the gate. And Grok has been just that, though Musk already had to rein the thing in last year when it started being a little too obviously white supremacist.
As of last week, Grok had limited these image-editing capabilities to paid subscribers of X, but as Ars Technica then reported, you could still do the same thing with the chatbot in its free desktop version, outside of X.
Now regulators in the UK, the European Union, and elsewhere are taking notice, particularly in light of the fact that there are many examples of sexualized image that feature what appear to be minors — some being more like anime, but many appearing more photo-realistic than that, as Wired reports.
UK regulators announced a probe of Grok and the deepfake situation on Monday, as Bloomberg reports, and this follows similar warnings and probes by the governments of France, India, and Indonesia. And Indonesia, along with Malaysia, became the first countries to fully block the Grok chatbot as of today.
Musk responded over the weekend to the growing outcry by calling the UK government "fascist" over its threats to block the X platform.
UK Business Secretary Peter Kyle said in a radio interview Monday that, because the UK has laws about sharing non-consensual sexual images, and against sharing sexualized images of children, it would "of course" consider blocking access to X altogether. And UK Prime Minister Keir Starmer's office has also put out a statement, per Bloomberg, saying that xAI's solution, to put the Grok image editor behind the subscription paywall, was insufficient.
The move, said a Starmer spokesperson, "simply turns an AI feature that allows the creation of unlawful images into a premium service: It’s not a solution." And, they added, "It’s insulting the victims of misogyny and sexual violence. What it does prove is that X can move swiftly when it wants to do so."
Musk's long-stated intention to create an AI model with few guardrails was sure to run up against some real-world troubles, and here is a case in point. But Musk has also long been a leader who prefers to act first and consider the consequences later, kind of like when he took a temp job with the Trump White House without considering what impact that might have on Tesla sales.
As of Monday, some of these sexualized images can still be seen in the Replies feed of Grok, though it seems like most of the requests are framed as coming from the women in the photo themselves.
In case you were wondering, Grok still treats "cis" as a slur, and while some of the overt antisemitism of last summer may have been toned down, it's not hard to find examples of racist and antisemitic tropes on the X platform — despite its ostensible policy against "hateful" conduct.
