Arizona’s attorney general filed criminal charges against prediction market Kalshi, accusing it of operating a gambling business without a license and offering illegal wagers on elections.

“Kalshi may brand itself as a ‘prediction market,’ but what it’s actually doing is running an illegal gambling operation and taking bets on Arizona elections, both of which violate Arizona law,” Attorney General Kris Mayes said in a statement on Tuesday.

While Arizona’s case is the first time criminal charges have been brought against the company, several other US states have alleged that Kalshi’s markets constitute illegal and unregulated sports betting.

“There’s clearly going to be a domino effect,” said Daniel Wallach, a lawyer who specializes in gaming law. “These are the first criminal charges filed against Kalshi anywhere in the US but they may not be the last.”

In a statement, Kalshi said: “Sadly, a state can file criminal charges on paper-thin arguments. States like Arizona want to individually regulate a nationwide financial exchange, and are trying every trick in the book to do it.”

Prediction market platforms such as Kalshi offer shares in binary outcomes, such as a certain team winning or losing a football match. Kalshi has argued that these contracts should continue to be regulated as derivatives by the federal Commodity Futures Trading Commission, enabling it to bypass states’ sports-gambling bans or regulations by arguing that its regulatory status under the CFTC preempts state-level laws.

https://arstechnica.com/tech-policy/2026/03/arizona-indicts-prediction-market-kalshi-for-running-illegal-gambling-operation/

The suit also alleges that this is part of a pattern of attacks on Colorado. Earlier in 2025, the US government moved a major Space Command facility from Colorado to Alabama. “When issuing his decision,” the suit continues, “the President stated that ‘the problem I have with Colorado’ is that ‘they do mail-in voting’ and that this ‘played a big factor’ in the decision.”

The suit notes that, on the same day as the NCAR announcement, the Department of Transportation killed $110 million in grants for projects in Colorado. Less than a week later, the Federal Emergency Management Agency rejected disaster relief requests from the state. And at the end of the month, Trump issued the first veto of his second term, rejecting a Colorado water management project. The suit presents these as indications that NCAR was just a casualty of this wide-ranging attack on a state that has opposed some of Trump’s agenda, and that the decision to close it was “untethered to any reasoned decision making.”

One thing that’s notable is that the suit does not even mention another potential motivation. NCAR is a major center of research on climate change, which the administration has repeatedly denigrated as a green scam, which would also have difficulty passing the reasoned decision-making standard. That said, there is a growing record of administration decisions that have been blocked based on the arbitrary and capricious standard. Courts have also been more willing than in Trump’s first term to look to public statements by administration figures (including Trump himself) as guides to the motivations behind decisions as well.

Regardless of how the case turns out, the discovery process is likely to provide a window into the actions of the head of the Office of Management and Budget, Russell Vought, who has largely stayed out of the public spotlight despite a controversial tenure in that office.

https://arstechnica.com/science/2026/03/university-group-sues-trump-administration-over-shutdown-of-climate-center/

Her first instinct was to contact the other victims she knew, then “ultimately, local law enforcement was contacted, and a criminal investigation was opened,” the complaint said.

Investigating the Discord evidence, cops quickly determined that the perpetrator had access to the first victim’s Instagram “because he had maintained a close and friendly relationship” with her. Searching his phone, cops found a third-party app that licensed or otherwise purchased access to Grok, which they concluded that the perpetrator used to morph the girls’ photos.

From there, the bad actor uploaded the images to a file-sharing platform called Mega and used them as a “bartering tool in Telegram group chats with hundreds of other users,” trading away the AI CSAM files “for sexually explicit content of other minors.”

The harms to victims have been extensive, the lawsuit said, citing acute emotional and mental distress. For the victims who know the perpetrator, they remain uncertain if the Grok-generated CSAM was shared with classmates or distributed to others at their school, the lawsuit noted. One girl fears the scandal will impact her college admissions, while another feels too scared to attend her own graduation.

Even more alarming than any acquaintances coming across the AI CSAM, however, is the fear that girls will now be stalked due to Grok’s outputs. As the lawsuit explains, “it also appears the victims’ true first names and the name of their school was attached to their files online, meaning other online predators may also be able to identify them, creating a substantial risk for stalking.”

xAI allegedly hosts Grok CSAM

While it was previously reported that Grok Imagine’s paying subscribers were generating more graphic outputs than the Grok outputs that sparked outcry on X, the lawsuit alleges that xAI has also taken other steps to hide how it profits from explicit content that harms real people.

https://arstechnica.com/tech-policy/2026/03/elon-musks-xai-sued-for-turning-three-girls-real-photos-into-ai-csam/

The attorneys general requested a response by March 2, one that addresses a set of leading questions, such as, “Why did the National Academies include a chapter on climate science that is not based on balanced or sound science?” and “What procedures will the National Academies establish to prevent similar advocacy-based chapters in future editions?” Since then, Ars has been contacting both the NAS and the Montana attorney general’s office (which published the letter) to try to find out whether a response was provided.

We finally learned yesterday that the response had been issued two days ahead of the deadline (it’s the final page of this PDF). Weighing in at just two sentences, the NAS says it used the same procedures to generate the climate chapter as it did for every other chapter, procedures that had been developed jointly with the Federal Judicial Center. “The manual, including the chapter on climate science, will continue to be available on the Academy’s website,” the response concludes.

The response leaves no obvious next step for the attorneys general. Their letter notes that the NAS is heavily dependent upon funding from the federal government to prepare its expert reports, and so producing reports that displease Republicans could be risky, but they have no ability to directly influence that funding.

Meanwhile, the political interference with the report drew a second response, this coming from many of the authors of the other chapters of the Reference Manual, who published an open letter decrying the political interference. In addition to noting the value of the Reference Manual and the rigorous peer review all chapters go through, the authors highlight the dangers posed by the actions of the attorneys general:

If political actors can determine which fields of established science are disfavored and off-limits to judicial education, every scientific discipline relevant to complex litigation becomes vulnerable to the same tactic. The integrity of the process by which judges evaluate scientific evidence should not be subject to political interference or veto.

The real danger is long term. If chapters continue to get deleted any time they run afoul of the current political winds, then it will become increasingly challenging to get the best scientists and legal scholars to contribute to the manual or its peer review. Over time, the quality of the material will decay, leaving judges less well-prepared to face cases with a heavy scientific component. Society as a whole will end up the loser.

https://arstechnica.com/science/2026/03/national-academies-of-sciences-resisting-pressure-to-pull-climate-info/

Trump’s quibble

Trump’s complaint said, “Yet again, an intentionally misleading headline by the Fake News Media about the five tanker planes that were supposedly struck down at an Airport in Saudi Arabia, and of no further use. In actuality, the Base was hit a few days ago, but the planes were not ‘struck’ or ‘destroyed.’ Four of the five had virtually no damage, and are already back in service. One had slightly more damage, but will be in the air shortly. None were destroyed, or close to that, as the Fake News said in headlines.”

The only specific news outlets Trump’s post mentioned were The New York Times and The Wall Street Journal, but he was referring to a Wall Street Journal article. Even if Trump’s version of events is true, his complaint wouldn’t meet the legal standard for proving a hoax or news distortion, or even prove that the Journal got anything wrong. Trump claims the planes were not “struck” but said four out of five “had virtually no damage,” which seemingly indicates that all five were struck and suffered some damage.

Trump’s post seems to accuse the Journal of falsely reporting that the planes were destroyed and would not be used again. But the Journal article makes it clear the planes were merely damaged, not destroyed, and would be repaired. It said:

Five US Air Force refueling planes were struck and damaged on the ground at Prince Sultan air base in Saudi Arabia, according to two US officials.

The tankers were hit during an Iranian missile strike on the Saudi base in recent days, the officials said. US Central Command declined to comment. The tankers were damaged but not fully destroyed and are being repaired, one of the officials said. No one was killed in the strikes.

The Journal article was published on Friday, and Trump issued his complaint on Truth Social on Saturday morning. The Journal article was updated on Saturday afternoon to include a quote from Trump’s Truth Social post. As far as we can tell, the article never claimed that any of the five tanker planes were destroyed. A Reuters article on Friday that quotes the WSJ report also uses the phrases “struck and damaged,” and “not fully destroyed,” undercutting Trump’s claim of false reporting.

https://arstechnica.com/tech-policy/2026/03/trump-and-his-fcc-chair-demand-more-positive-news-coverage-of-iran-war/

ChatGPT’s “adult” filters have been buggy

Sources told the WSJ that they doubted if OpenAI’s tools were ready to lock kids out of prohibited content.

Their whistleblowing comes after OpenAI fired a top safety executive who opposed the release of “adult mode.” OpenAI denied the firing was related, but the exiting staffer directly criticized both the AI firm’s ability to block kids from content and stop outputs from promoting child exploitation. Further, a second former safety staffer also spoke out last fall, warning that parents shouldn’t trust OpenAI’s “adult mode” claims.

To counter this narrative, OpenAI’s spokesperson promised that the company “has a developed plan to monitor for a range of potential long-term effects of adult mode, both positive and negative.”

However, that plan was likely developed with the very experts the WSJ reported are staunchly opposing the roll-out, leaving parents to wonder if OpenAI cares about advice from its youth well-being team or not.

On top of ineffective age checks or clever minors who dodge age gates, OpenAI may get in trouble with parents if its own systems unexpectedly fail. Back in April when OpenAI started dabbling with more risqué outputs, OpenAI fixed a bug that TechCrunch testing found was allowing minors to access graphic erotica on ChatGPT. It seems that OpenAI’s filters broke that were supposed to clearly restrict “sensitive content like erotica to narrow contexts such as scientific, historical, or news reporting.”

“In this case, a bug allowed responses outside those guidelines, and we are actively deploying a fix to limit these generations,” OpenAI said at the time.

OpenAI did not respond to Ars’ request to comment.

If you or someone you know is feeling suicidal or in distress, please call the Suicide Prevention Lifeline number by dialing 988, which will put you in touch with a local crisis center.

https://arstechnica.com/tech-policy/2026/03/chatgpt-may-soon-become-sexy-suicide-coach-openai-advisor-reportedly-warned/

Over 7 million cable and fiber locations

Astound is already the product of industry consolidation via a series of private equity deals that combined Wave Broadband, RCN, and Grande Communications. A research note from the New Street analyst firm said GFiber offers service at 2.8 million locations in 15 states, while Astound’s service area has 4.45 million locations in 12 states and the District of Columbia. Most of Astound’s network is cable broadband, but it has 892,014 fiber locations and 44,548 copper locations.

“Put together, the two companies pass ~7.1 [million] locations in 26 states,” the research note said. “The two companies overlap in only three counties in Texas (109k locations). Texas and Illinois will have the largest footprint for the combined entity. Cable and Fiber will cover an almost equal share of locations for the combined company.”

The combined GFiber/Astound company will face competition in most of its territory from at least one cable or fiber/copper provider. That includes AT&T at 53 percent of locations, Comcast at 46 percent of locations, Charter at 43 percent of locations, Verizon at 22 percent of locations, and Lumen (CenturyLink) at 11 percent.

New Street said there are unanswered questions, such as whether the combined company will continue to expand into areas served by existing cable and fiber operators, and whether it will upgrade its own cable footprint with fiber.

GFiber offers service in Texas, North Carolina, Missouri, Utah, Kansas, Tennessee, Alabama, Georgia, Iowa, California, Arizona, Nebraska, Idaho, Colorado, and South Carolina. Astound is in Illinois, Texas, New York, California, Washington, Massachusetts, Pennsylvania, the District of Columbia, Oregon, Maryland, Indiana, Virginia, and New Jersey.

Texas, North Carolina, Missouri, Utah, and Kansas account for about 78 percent of GFiber locations, according to New Street’s data. Illinois, Texas, New York, California, and Washington together account for about 72 percent of Astound’s locations.

https://arstechnica.com/tech-policy/2026/03/google-fiber-will-be-sold-to-private-equity-firm-and-merge-with-cable-company/

If you’re an AT&T FirstNet customer and suddenly get hit with a $6,200 charge, the good news is that it’s probably a mistake and can be corrected. But actually getting the wrong charge wiped out might not be so easy.

This has now happened at least twice. In December 2024, a Texas police officer received a $6,223 bill with a $6,194 charge for using 3.1GB of data. He said he had unlimited data but was charged incorrectly after moving a line to AT&T’s FirstNet service for first responders. He called AT&T and went to an AT&T store but only got the bill reversed after contacting the AT&T president’s office.

An AT&T spokesperson told Ars at the time that it was “investigating to determine what caused this system error.” But AT&T never revealed exactly what caused it, and now another FirstNet user has gone through an almost identical ordeal.

On Monday this week, an active-duty military member from Florida emailed us and said he was “experiencing the exact same issue with AT&T.” The man, who preferred that we not publish his name, showed us a bill with a $6,196 charge for about 3.1GB of data use.

$2 per megabyte

The bill stated he had FirstNet Unlimited service from January 19 to February 14. There was also a note of a service change on January 19 and a charge for “FN Data PPU 3,098MB at $2.00 per MB,” adding up to $6,196. This was almost identical to the Texas man’s December 2024 bill that had a $6,194 line item listed as “Data Pay Per use 3,097MB at $2.00 per MB.”

We contacted AT&T on Tuesday with details about the Florida military member’s bill. We asked AT&T if it ever figured out what caused the similar billing problem in December 2024, whether it made any changes to prevent it from happening again, and for details on the cause of the new error.

AT&T didn’t answer our specific questions, but it fixed the bill within a couple of hours. “This was a big relief,” the FirstNet customer told us later that day, saying that AT&T “knocked it out completely today off of my bill.”

https://arstechnica.com/tech-policy/2026/03/att-wrongly-charged-man-6196-reversed-bill-after-hearing-from-ars/

Weinhold wrote in another chat, “I have VIP parking up to $250 lol.” Baker replied, “I almost feel bad taking advantage of them.” Weinhold then mentioned that he raised club prices to $125, and Baker replied, “I wonder if I can get $225.”

Live Nation said the messages aren’t reflective of the company’s general operations. “The Slack exchange from one junior staffer to a friend absolutely doesn’t reflect our values or how we operate,” Live Nation said in a statement provided to Ars today. “Because this was a private Slack message, leadership learned of this when the public did, and will be looking into the matter promptly. Our business only works when fans have great experiences, which is why we’ve capped amphitheater venue fees at 15 percent and have invested $1 billion in the last 18 months into US venues and fan amenities.”

The US and states said Live Nation is downplaying Baker’s position at the company. “Defendants’ brief fails to mention this individual has since been promoted and now serves as Head of Ticketing for Venue Nation, with responsibilities relating to all of Live Nation’s venues,” the plaintiffs’ brief said.

Live Nation said in a March 8 filing that the messages aren’t relevant to the trial because they concerned fees for things like VIP club access, premier parking, or lawn chair rentals. “These products are not primary concert tickets, are sold separately from tickets, and are not part of the ticketing services markets at issue in this trial; they bear no relevance to the parties’ claims and defenses,” Live Nation told the court.

Live Nation: Messages could “inflame the jury”

Live Nation said the only purpose of using the exhibits as evidence “is to portray Defendants in an unflattering light and inflame the jury against Defendants,” and that the exhibits “would confuse and mislead the jury, invite decision-making on an improper emotional basis, and cause unfair prejudice to Defendants.” The company also asked the court to bar plaintiffs “from questioning Ben Baker or any other witness about the substance of these Exhibits or about similar communications concerning ancillary, fan-facing products and services not encompassed by the markets and claims proceeding to trial.”

https://arstechnica.com/tech-policy/2026/03/live-nation-director-boasted-of-gouging-ticket-buyers-robbing-them-blind/

Bankman-Fried’s plan to reinvent himself as a Republican, Buckley noted, was detailed in a Google Document that the court reviewed before convicting Bankman-Fried in 2024.

Buckley said the document showed how, “in the aftermath of FTX’s collapse,” Bankman-Fried “mapped out a rehabilitation and pardon campaign.” Attached to an email from Bankman-Fried’s account, the Google Doc was marked “confidential” and started with a note that emphasized that “these are all random probably bad ideas that aren’t vetted.”

However, many of the ideas were executed as planned, Buckley wrote. For example, Bankman-Fried planned to “come out as Republican” in an interview with Tucker Carlson, which happened.

“In March 2025, the defendant gave an interview to Tucker Carlson in which he portrayed himself as a disaffected Democrat who had become sympathetic to Republicans before his arrest” and “suggested his political reorientation contributed to his prosecution,” Buckley wrote.

Bankman-Fried also, in his document, considered using X to “come out against the woke agenda” and push the narrative that he had hidden Republican donations, which also happened.

“That checklist is being executed with near-perfect fidelity,” Buckley alleged. However, the plan isn’t working, and Bankman-Fried’s X posts aren’t causing Trump officials to warm to him, he said. “Evidence, not politics, drove the Government’s prosecution of the defendant,” Buckley insisted.

“Contrary to his claim that he has been targeted for his politics, the public record establishes unambiguously that the defendant was a major, publicly identified financial supporter of Democratic causes,” Buckley wrote. Later, he emphasized, “The motion’s suggestion that he was somehow prosecuted because of his party affiliation inverts the factual reality: he was a major donor, not a political adversary.”

https://arstechnica.com/tech-policy/2026/03/trumps-doj-is-not-falling-for-sam-bankman-frieds-maga-makeover-on-x/