Elon Musk has quietly bought APR Energy, a Jacksonville-based company that operates a fleet of mobile gas and diesel turbines totaling more than 1 GW of generation capacity.
The self-styled champion of a “solar electric economy” now owns a fossil fuel power company — and he’s buying it to feed the electricity-hungry data centers running xAI’s Grok.
It should come as no surprise, then, that its supposed “translation” is a piece of work, too. In April, the almost-trillionaire’s social media platform X instated automatic AI translations for all of its users — and the results certainly speak for themselves.
As writer and author Parker Molloy pointed out in a recent post on Bluesky, the Grok feature is “taking some interesting liberties” with people’s otherwise sincere posts.
Screenshots show how Grok completely botched translations by coming up with shocking and decidedly NSFW AI hallucinations.
One South Korean user, for instance, posted an otherwise tame video of two video game characters, only for the auto-translate feature to allege that it was a “c**shot video with my stepmom.”
“I am not sure it is the right translation,” a taken-aback user replied.
“It isn’t correct at all!” the original poster lamented.
Another post by Portuguese account POPTime shows a video of a man making an elaborate cup of coffee during a flight.
But a purported translation “from Portuguese by Grok” claimed that the video shows a man who is pleasuring himself in public.
“Grok auto translate is inaccurate, text actually translates to ‘Man grinds and brews his own coffee during a commercial flight,’ not ‘Man masturbates and jerks off to his own coffee during commercial flight,'” an appended community note reads.
Another Turkish user, who posted an innocent picture of their young kitten, was horrified after Grok’s translation suggested that they were looking to “f*** our baby.”
“Uhhh the automatic translation makes it alarming,” another user replied.
To the X community, which has already seen its fair share of Grok’s foul-mouthed shenanigans, the AI’s profanity-laden translations were just another bemusing distraction.
But to anybody looking to meaningfully post on social media, the bizarre liberties the chatbot is taking perfectly highlight the sorry state of the microblogging platform. Ever since Musk took over in 2022, users have noted a major uptick in the proliferation of hate speech and disinformation. Researchers have found that X’s AI models are driving online racist abuse.
Grok’s ability to come up with AI images, in particular, has been mired in controversy, from being used to generate nonconsensual nudes of women to a single account that was recently found to have posted thousands of explicit images of an 11-year-old girl.
Musk’s prodigal chatbot, in other words, has been guilty of crimes far worse than expletive-laden translations.
According to boosters of AI, the tech was supposed to usher in an era of unprecedented productivity, leisure time for workers, and scientific discovery. But in reality, it keeps being linked to crimes so ghoulish that they’re almost unbelievable.
In a particularly horrifying example, a man used Elon Musk’s Grok chatbot to generate some 7,000 sexually explicit images of his stepdaughter by feeding the AI a photo of her that was taken when she was 11 years old, a lawsuit alleges.
The allegations, as reported by Ars Technica, were made in a proposed class action lawsuit against xAI — now owned by SpaceX and rebranded as SpaceXAI— which accuses it of building AI tools to “nudify” people without their consent and shielding predators from law enforcement.
According to the suit, brought by a group of girls who say they were victimized by obscene Grok generations, the stepfather’s horrific activity almost went entirely unnoticed by Grok’s safety systems. It failed to flag the countless images depicting incest and rape, per Ars. It was only after a specific prompt for a “gang rape” that it sent a tip to the National Center for Missing and Exploited Children (NCMEC), who alerted authorities.
But xAI “obstructed” the investigation that followed “at every turn,” the suit alleges. It refused to provide more information on the user behind the requests, including critical identifying details like an IP address. Police tracked down the man anyway, finding a vast collection of AI-generated images of his stepdaughter, identified only as Jane Doe 4. Two days after being released on bail, the man fatally shot himself, upending the young girl’s life even more than he already had.
“Overnight, Jane Doe 4’s entire reality was shattered by the dual tragedies of child sexual exploitation and suicide,” the lawsuit alleged. “Her family was torn apart, and her life became a nightmare.”
Grok’s ability to create sexual deepfakes gained national attention after a several week period in last December and this January when it was used to generate tens of thousands of nonconsensual AI nudes of real women and children. In the aftermath, the firm Copyleaks estimated the bot was generating a nonconsensually sexualized image every single minute. The Center for Counter Digital Hate later estimated that this amounted to around 3,000,000 AI nudes, with more than 23,000 of those images depicting children.
Musk, the CEO and owner of xAI, has denied that Grok has been used to generate CSAM. The company’s immediate response to the outcry was to simply restrict image generation to paying users. Over a week later — as Grok continued to crank out more CSAM and deepfake nudes — xAI said it was implementing stricter guardrails to stop Grok from generating sexual images of real people, without acknowledging the disturbing trend.
The lawsuit also accuses xAI of protecting predators by frustrating law enforcement’s attempts to track them down. In a press release quoted by Ars, the plaintiff’s legal team pointed to NCMEC’s discovery that 90 percent of xAI’s CyberTips weren’t “actionable” because they barely provided information on the flagged users.
“They had everything they needed to help law enforcement stop the person responsible and achieve justice. Instead, they remained silent and allowed this person to use Grok to steal my childhood,” Jane Doe 4 said, per Ars.
Elon Musk told Tesla staff to move to using Grok, the AI model from his own xAI (now folded into SpaceX), according to a memo sent to employees on Friday.
The push comes days after Tesla capped employee spending on third-party AI tools — and it lands even though Musk himself concedes Grok is not as good as its rivals.
Login
