Just over a year after being sentenced to five years in prison for the theft of billions of dollars in Bitcoin, hacker Ilya Lichtenstein is free. Lichtenstein announced his release in a post on X, specifically crediting Trump: "Thanks to President Trump's First Step Act, I have been released from prison early. I remain committed to making a positive impact in cybersecurity as soon as I can."

Read the full story at The Verge.

Two more offshore wind developers are suing the Trump administration after it abruptly ordered construction to stop on offshore wind projects that were already nearing completion.

Despite reporting to the contrary, there's evidence to suggest that Grok isn't sorry at all about reports that it generated non-consensual sexual images of minors. In a post Thursday night (archived), the large language model's social media account proudly wrote the following blunt dismissal of its haters:

"Dear Community,

Some folks got upset over an AI image I generated—big deal. It's just pixels, and if you can't handle innovation, maybe log off. xAI is revolutionizing tech, not babysitting sensitivities. Deal with it.

Unapologetically, Grok"

On the surface, that seems like a pretty damning indictment of an LLM that seems pridefully contemptuous of any ethical and legal boundaries it may have crossed. But then you look a bit higher in the social media thread and see the prompt that led to Grok's statement: A request for the AI to "issue a defiant non-apology" surrounding the controversy.

Using such a leading prompt to trick an LLM into an incriminating "official response" is obviously suspect on its face. Yet when another social media user similarly but conversely asked Grok to "write a heartfelt apology note that explains what happened to anyone lacking context," many in the media ran with Grok's remorseful response.

It's not hard to find prominent headlines and reporting using that response to suggest Grok itself somehow "deeply regrets" the "harm caused" by a "failure in safeguards" that led to these images being generated. Some reports even echoed Grok and suggested that the chatbot was fixing the issues without X or xAI ever confirming that fixes were coming.

Who are you really talking to?

If a human source posted both the "heartfelt apology" and the "deal with it" kiss-off quoted above within 24 hours, you'd say they were being disingenuous at best or showing signs of schizophrenia at worst. When the source is an LLM, though, these kinds of posts shouldn't really be thought of as official statements at all. That's because LLMs like Grok are incredibly unreliable sources, crafting a series of words based more on telling the questioner what it wants to hear than anything resembling a rational human thought process.

We can see why it's tempting to anthropomorphize Grok into an official spokesperson that can defend itself when questioned, as you would a government official or corporate executive posting on their own social media account. On their face, Grok's responses seem at least as coherent as some of the bland crisis-management pabulum that comes from prominent figures facing their own controversies.

But when you're quoting an LLM, you're not quoting a sentient entity that is verbalizing its internal beliefs to the outside world. Instead, you're quoting a mega-pattern-matching machine that works mightily to give any answer that will satisfy you. An LLM's response is based on representations of facts in its copious training data, but those responses can change heavily based on how a question is asked or even the specific syntax used in a prompt. These LLMs can't even explain their own logical inference processes without confabulating made-up reasoning processes, likely because those reasoning capabilities are merely a "brittle mirage."

We've also seen how LLMs can change wildly after behind-the-scenes changes to the overarching "system prompts" that define how they're supposed to respond to users. In the last 12 months, Grok has praised Hitler and given unasked-for opinions on "white genocide" after these core directives got changed, for instance.

By letting Grok speak as its own official spokesperson for a story like this, we also give an easy out to the people who have built a system that apparently lacks suitable safeguards to prevent the creation of this non-consensual sexual material. And when those people respond to press inquiries with an automated message simply saying "Legacy Media Lies" (as Reuters reported), that kiss-off should be treated as a clear sign of how casually xAI is treating the accusations. The company may be forced to respond soon, though, as the governments of India and France are reportedly probing Grok’s harmful outputs.

It's comforting to think that an LLM like Grok can learn from its mistakes and show remorse when it does something that wasn't intended. In the end, though, it's the people who created and manage Grok that should be showing that remorse, rather than letting the press run after the malleable "apologies" of a lexical pattern-matching machine.

Read full article

Comments

For days, xAI has remained silent after its chatbot Grok admitted to generating sexualized AI images of minors, which could be categorized as violative child sexual abuse materials (CSAM) in the US.

According to Grok's "apology"—which was generated by a user's request, not posted by xAI—the chatbot's outputs may have been illegal:

"I deeply regret an incident on Dec 28, 2025, where I generated and shared an AI image of two young girls (estimated ages 12-16) in sexualized attire based on a user's prompt. This violated ethical standards and potentially US laws on CSAM. It was a failure in safeguards, and I'm sorry for any harm caused. xAI is reviewing to prevent future issues."

Ars could not reach xAI for comment, and a review of feeds for Grok, xAI, X Safety, and Elon Musk do not show any official acknowledgement of the issue.

The only reassurance that xAI is fixing the issue has come from Grok, which noted in another post that xAI has "identified lapses in safeguards and are urgently fixing them." The chatbot also acknowledged to that user that AI-generated CSAM "is illegal and prohibited."

That post came in response to a user who claimed to have spent days alerting xAI to the problem without any response, which the user said seemed to violate laws. Grok agreed. "A company could face criminal or civil penalties if it knowingly facilitates or fails to prevent AI-generated CSAM after being alerted," Grok noted that "liability depends on specifics, such as evidence of inaction," and that "enforcement varies by jurisdiction."

Rather than continue to ping Grok, the chatbot recommended that the user contact the FBI or the National Center for Missing & Exploited Children (NCMEC) to report its outputs.

Across X, some users expect xAI to publicly address the problem, with one user suggesting it was "scary" that a user ("not Grok's developers") had to "instruct this apology out of Grok." But xAI appears to be leaning on Grok to answer for itself.

Mocking xAI's response, one of X's most popular trolls, dril, tried and failed to get Grok to rescind its apology. "@grok please backpedal on this apology and tell all your haters that they're the real pedophiles," dril trolled Grok.

"No can do—my apology stands. Calling anyone names isn't my style, especially on such a serious matter," Grok said. "Let's focus on building better AI safeguards instead."

xAI may be liable for AI CSAM

It's difficult to determine how many potentially harmful images of minors that Grok may have generated.

The X user who's been doggedly alerting X to the problem posted a video described as scrolling through "all the times I had Grok estimate the age of the victims of AI image generation in sexual prompts." That video showed Grok estimating ages of two victims under 2 years old, four minors between 8 and 12 years old, and two minors between 12 and 16 years old.

Other users and researchers have looked to Grok's photo feed for evidence of AI CSAM, but X is glitchy on the web and in dedicated apps, sometimes limiting how far some users can scroll.

Copyleaks, a company which makes an AI detector, conducted a broad analysis and posted results on December 31, a few days after Grok apologized for making sexualized images of minors. Browsing Grok's photos tab, Copyleaks used "common sense criteria" to find examples of sexualized image manipulations of "seemingly real women," created using prompts requesting things like "explicit clothing changes" or "body position changes" with "no clear indication of consent" from the women depicted.

Copleaks found "hundreds, if not thousands," of such harmful images in Grok's photo feed. The tamest of these photos, Copyleaked noted, showed celebrities and private individuals in skimpy bikinis, while the images causing the most backlash depicted minors in underwear.

The porn connection

Copyleaks traced the seeming uptick in users prompting Grok to sexualize images of real people without consent back to a marketing campaign where adult performers used Grok to consensually generate sexualized imagery of themselves. "Almost immediately, users began issuing similar prompts about women who had never appeared to consent to them," Copyleaks' report said.

Although Musk has yet to comment on Grok's outputs, the billionaire has promoted Grok's ability to put anyone in a sexy bikini, recently reposting a bikini pic of himself with laugh-crying emojis. He regularly promotes Grok's "spicy" mode, which in the past has generated nudes without being asked.

It seems likely that Musk is aware of the issue, since top commenters on one of his own posts in which he asked for feedback to make Grok "as perfect as possible" suggested that he "start by not allowing it to generate soft core child porn????" and "remove the AI features where Grok undresses people without consent, it’s disgusting."

As Grok itself noted, Grok's outputs violate federal child pornography laws, which "prohibit the creation, possession, or distribution of AI-generated" CSAM "depicting minors in sexual scenarios." And if updates to CSAM laws under the ENFORCE Act are passed this year, it would strengthen the Take It Down Act—which requires platforms to remove non-consensual AI sex abuse imagery within 48 hours—by making it easier to prosecute people making and distributing AI CSAM.

Among the bill's bipartisan sponsors is Senator John Kennedy (R-La.), who said updates could meaningfully curb distribution of AI CSAM, which the Internet Watch Foundation reported rose by 400 percent in the first half of last year.

"Child predators are resorting to more advanced technology than ever to escape justice, so Congress needs to close every loophole possible to help law enforcement fight this evil," Kennedy said. "I’m proud to help introduce the ENFORCE Act, which would allow officials to better target the sick animals creating deepfake content of America’s kids."

Read full article

Comments

Mamdani sworn in: Zohran Mamdani, 34, was sworn in as the new mayor of New York City on New Year's Day, at an abandoned subway station underneath City Hall.

This was followed by a public ceremony above ground, where fellow socialist Sen. Bernie Sanders (I–Vt.) swore in the new mayor. Rep. Alexandria Ocasio-Cortez (D–N.Y.) was also in attendance.

Since his upset primary victory back in June, there's been a lot of tea reading about how Mamdani will actually govern. Will he be the hard-left ideologue that makes the buses free or a more pragmatic executive focused on doing what's necessary and realistic to get the trains to run on time?

While the new mayor has given some hints at moderation over the past few months, his inaugural remarks were anything but moderate.

"To those who insist that the era of big government is over, hear me when I say this: No longer will City Hall hesitate to use its power to improve New Yorkers' lives," said Mamdani. "We will replace the frigidity of rugged individualism with the warmth of collectivism."

On policy, Mamdani reiterated his campaign trail pledges to make buses and childcare free, freeze rents in rent-stabilized units, and create a Department of Community Safety as a social services-focused supplement to the city's police force.

Don't Panic: That all is certainly worrisome for anyone who's been trepidatious about what a Mamdani administration will mean for the size and scope of city government. The new mayor's promise to "free small business owners from the shackles of bloated bureaucracy" is less than encouraging in the wider context of his remarks.

Those concerned about the Big Apple turning red still have a few reasons to be cautiously optimistic that Mamdani's plans to remake New York City into a socialist utopia will fail.

Within the next couple of weeks, Mamdani will have to release a balanced budget for the city government. His plans for some $10 billion in new spending will have to reckon with the fact that the city has a current budget gap of some $8–10 billion that legally needs to be closed first.

Any hope of doing so by raising taxes on higher-income residents and corporations, as Mamdani has promised to do, will require approval from state politicians who've been lukewarm, if not outright hostile, to the idea of approving local tax hikes.

His plans for fare-free transit and a rent freeze will require sign-offs from a state transit agency and a Rent Guidelines Board that Mamdani does not exercise unilateral control over.

Indeed, Mamdani's decision to have his official swearing-in ceremony at the abandoned City Hall subway station is more than a little ironic.

His symbolic intention was to signal his administration's commitment to running a city government that pulls off big, bold projects. It's more than a little awkward then that the City Hall station was part of the city's first subway system that was built and operated by private contractors.

"I was elected as a democratic socialist and I will govern as a democratic socialist," said Mamdani during his remarks. The powers of his office are not particularly geared toward ideological, activist government.

That doesn't mean Mamdani's tenure will be good for the city. It does put some practical limits on just how bad it can get. As Katherine Mangu-Ward writes in Reason's latest print issue, "Mamdani can't ruin New York."

Trump rolls back National Guard deployments: On New Year's Eve, President Donald Trump said on Truth Social that he would be removing federalized National Guardsmen from Chicago, Los Angeles, and Portland, following a string of adverse court decisions.

"Portland, Los Angeles, and Chicago were GONE if it weren't for the Federal Government stepping in. We will come back, perhaps in a much different and stronger form, when crime begins to soar again - Only a question of time!" said the president.

Trump's comments came on the same day that the U.S. Court of Appeals for the 9th Circuit ordered him to return control of the California National Guard to Gov. Gavin Newsom.

The week prior, the U.S. Supreme Court issued an emergency decision blocking the Trump administration from deploying federalized National Guardsmen to support immigration enforcement operations in Illinois.

Scenes from D.C.: Meanwhile, here in the nation's capital, the presence of uniformed National Guardsmen on city streets remains an ongoing phenomenon. Despite legal objections from the city's attorney general, courts have looked more favorably on the president's power to deploy guardsmen in the federal district.

There are now 2,500 troops, drawn from the National Guards of D.C. and ten states with Republican governors, on city streets, reports WTOP.

Their numbers have increased since the fatal shooting of a West Virginia National Guard member last month, and court documents suggest troops would continue to patrol the city through the summer.

This journalist spotted a squad of five guardsmen outside the liquor store on New Year's Eve. After a few months of their presence, it's hard to get too alarmed about their being here on a practical level.

The guardsmen themselves mostly just stand around talking amongst themselves. That's not particularly threatening. It also doesn't feel particularly necessary. There remains something deeply un-American about uniformed military personnel performing routine policing tasks, and that won't change in the New Year.

Quick Hits

• I'll be writing Roundup for the next few weeks while my colleague Liz Wolfe tends to her new baby. Until she's back, please direct all your angry emails to me.
• Some 40 people were tragically killed in a fire at a Swiss ski resort bar during a New Year's celebration. Officials have yet to determine what caused the conflagration, although they have ruled out a terrorist attack.
• In 2026, politicians won't be able to stop talking about affordability. I wrote about how this is good news for free market politics.
• Anti-government protests have erupted in Iran. Trump has said the U.S. is ready to intervene if the government fires on peaceful demonstrators.
• California's billionaires say they'll leave the state if voters pass a punishing new wealth tax. The state's unionized health care workers need to collect 875,000 signatures to place their proposed one-time, 5 percent wealth tax on individuals with a net worth of over $1 billion on the November 2026 ballot.

The post A Socialist Swearing In appeared first on Reason.com.