Those of us missing the John Wick universe can take heart, because Chad Stahelski—who created that franchise—is coming back early next year with a new dystopian action film. It’s called Karōshi (a Japanese term that means “death overwork”), and Lionsgate released a killer trailer that puts this film at the top of my list of “must see” movies of 2027.

Dying from working too hard is a very real phenomenon, not just in Japan and much of Asia, but also in Sweden, where some 700 workers die each year from overwork. The physical cause is usually stress-related heart attacks or strokes, as well as suicide (karōjisatsu in Japan). Writer/director Takashi Doscher has transported that concept to a grim, futuristic corporate Manhattan, specifically a company called Ikigai Technologies. Per the official premise:

Karōshi is an action-thriller set in a near-future Manhattan, where corporate loyalty and cultural tradition have reshaped everyday life. When a mysterious outsider infiltrates a powerful company after a worker’s death, he exposes the violent system hidden beneath its polished surface.

Teo Yoo (Past Lives) stars as the aptly named Hiro, who is clearly out for vengeance against the offending corporation. The cast also includes Takehiro Hira, so memorable in the 2024 miniseries Shōgun, as well as Giancarlo Esposito, Cynthia Erivo, Bill Camp, Richard Harmon, Isabel May, and Show Kasamatsu.

The trailer opens with Hira’s character lauding the contribution of Ikigai Technologies in helping shape “modern life for millions,” adding praise for employees’ “daily commitment and sacrifice.” Then we see one distraught worker deliberately falling backward to his death from the top of the company’s skyscraper. Enter our protagonist, Hiro, who suspects something more sinister is going on behind the company’s embrace of the samurai path, marked by “integrity” and “honor”—and also violence.

It’s not long before Hiro is battling all kinds of opponents in fine Wickian fashion, including using a paper umbrella to shield himself from the blood spurting from the neck of an opponent he’s just killed—a clever flourish. John Wick would be proud.

Karōshi hits theaters on February 5, 2027.

There are optimistic plans for spaceflight, and then there are optimistic plans.

AST SpaceMobile is a Texas-based, publicly traded company building very large satellites to provide direct-to-cell communications on Earth. But it has been having trouble finding rides to space. So on Friday, AST announced that it is developing a medium-lift rocket, named Blacklift, which will have the capacity to lift 20 metric tons to low-Earth orbit. Oh, and the company said it would start launching commercially in late 2028.

“The Blacklift rocket is the springboard to significantly expand our total addressable market, providing AST SpaceMobile with regular unconstrained access to orbit and providing for most favorable owner’s economics in the industry,” said Abel Avellan, founder, chairman, and CEO of AST SpaceMobile, in a news release.

To anyone with even a basic understanding of rockets, that timeline is wildly, fantastically, and stupidly optimistic. Still, the announcement is worth examining because it underscores two important trends in today’s launch industry.

First, it’s becoming increasingly apparent that satellite operators planning to deploy large constellations need to secure launch capacity years in advance, as Amazon did nearly half a decade ago. And if they can’t secure that capacity on favorable terms, they are now considering taking the extraordinary step of developing their own rockets.

And secondly, since investors are now recognizing the worsening launch crisis that Ars has been writing about for months, there is apparently capital to be raised for new entrants into the medium-lift launch market. AST is not the only company making fantastical claims of late. A firm called Cowboy Space said in October it was building a large rocket named Big Iron (full credit—that’s an excellent name) to launch orbital data centers. It also claimed its large rocket would make its first flight in 2028.

Sure thing, y’all.

Comparable launch development timelines

Let’s start with a little history. SpaceX became the first purely commercial launch startup to announce plans to build a new medium-lift rocket about two decades ago, in 2005. The company holds the record for the fastest time to orbit with the Falcon 9 vehicle, with a development time of 4.7 years.

But SpaceX had some significant advantages. By then, the company had been operating for more than three years, had extensive experience with the Merlin engine, and was less than a year away from its first launch of the smaller Falcon 1 rocket. In retrospect, it’s also clear that the company had an insanely talented team of engineers. Finally, the Falcon 9 was quite a bit smaller than the recently announced vehicles, Blacklift and Big Iron. In its initial form, Falcon 9 had a lift capacity of about 10 tons.

What about other companies? Since 2015, a number of other US companies—United Launch Alliance (Vulcan), Rocket Lab (Neutron), Relativity Space (Terran R), Firefly/Northrop Grumman (Eclipse), and Stoke Space (Nova) have announced medium to heavy lift vehicles, ranging in payload capacity from about 7 to 27 metric tons (Blue Origin’s New Glenn is much larger than any of these).

Of these, only Vulcan has reached orbit. It took 8.7 years, with an incredibly experienced launch team behind the vehicle and about $2 billion in aggregate federal support across its suppliers.

Broadly speaking, the track records of these existing launch companies suggest that developing a new medium- or heavy-lift rocket takes five to nine years from public announcement to first orbital launch. Moreover, many of these companies either started with smaller rockets or brought other valuable experience to bear.

Yes, they’re getting some engines—at a cost

AST SpaceMobile’s counterargument is that it won’t be starting from scratch. The company said Friday that it has signed an agreement with Firefly to deliver engines for the first three Blacklift rockets, provide engineering services over a three-year period, and acquire the perpetual license to manufacture the Miranda and vacuum-variant Vira engines from Firefly. Nine Miranda engines will power Blacklift’s first stage.

Yes, engines are typically the hardest part of a rocket’s development. And procuring Miranda engines will shave some time off Blacklift’s development. But it will probably be a year or two, not five. Procuring the Miranda engine—which likely remains at least two years from being tested in flight on the Eclipse rocket—does not eliminate the enormous engineering, manufacturing, testing, and regulatory effort required to develop a medium-lift rocket.

AST SpaceMobile is also paying a pretty penny for this privilege. It has committed $1.3 billion to Firefly for the engine license ($800 million), 27 Miranda engines and nine Vira engines ($250 million), and engineering support ($250 million). If we look just at the engine costs for a single Blacklift rocket, that’s $83 million per vehicle for the first three.

The bottom line is that AST SpaceMobile has made a launch timeline claim that is profoundly disconnected from the historical record of every comparable rocket development program this century. Maybe they’re a bunch of geniuses who will vibe-code the vehicle with Claude or Grok and shock us all.

More likely, though, is that this is a company that desperately needs to launch about 90 more of its massive BlueBird direct-to-cell satellites. Yes, the company has some existing commercial launch contracts with SpaceX, Blue Origin, and other companies. And in its news release, it said, “AST SpaceMobile will continue to use external providers, with agreements now covering in total over 100 BlueBird satellites to be deployed cumulatively between 2027 and 2028.”

But that raises an obvious question: If AST already has agreements covering the deployment of more than 100 BlueBird satellites by the end of 2028, what precisely is Blacklift supposed to accomplish? Do you spend several billion dollars to develop a rocket you don’t actually need?

After weeks of backlash—and a failed attempt to quell it—Meta has indefinitely delayed plans to update its commercial terms and require Meta Business account users to allow extensive AI training on external websites linked on Facebook.

“We’re pausing the ‌changes ⁠as we continue to engage with stakeholders and ​take ​our ⁠time to get this right. ​In the ​meantime, ⁠our existing terms remain in ⁠effect,” a Meta spokesperson told Reuters.

Soon after Meta notified business account holders of the upcoming changes, legal experts warned Nieman Lab that the vague language could allow the company to ignore scraping preferences specified in robots.txt files and exhaustively crawl vast amounts of off-Facebook content for AI training. Since virtually every news publisher has a Facebook page, Nieman Lab sounded the alarm that Meta could potentially avoid millions in licensing fees if the TOS update happened as planned.

Ars broke down the changes after Meta announced them, noting that businesses—including news publishers involved in drawn-out legal battles over unwanted AI scraping—had no way to opt out. To keep a business Facebook page, they had to allow Meta to crawl any linked external sites. The terms even required publishers to create test credentials so Meta’s bots could reliably bypass paywalls.

Meta faced global scrutiny

The changes were initially supposed to take effect on October 30. But Meta pushed that date back to November 20 amid pushback from Facebook users, courts, and regulators around the world. Courts in Brazil and Norway issued preliminary injunctions blocking the terms after organizations in those countries promptly filed lawsuits, Nieman Lab reported.

The United Kingdom’s Competition and Markets Authority also opened a formal investigation into whether Meta was exploiting a dominant position to seize valuable AI training data. In a LinkedIn post, Theo Bamber, chief executive of the UK’s News Media Association, pointed to “the continued arrogance of this company to think everyone will just be OK with its actions” as “staggering.”

“Does Meta think we can’t read?” Bamber wrote. He argued that removing the word “train” from the terms of service did little to reassure publishers, since “develop” posed the same threat.

“It’s saying it can use all the content it scrapes to develop ‘any business product’. I mean, come on,” he wrote.

Nieman Lab noted there was bipartisan concern in the US. Sens. Adam Schiff (D-Calif.) and Marsha Blackburn (R-Tenn.) sent letters asking Mark Zuckerberg to withdraw the terms.

On October 28, Meta’s vice president of communications, Andy Stone, posted on X that Meta would do just that, at least for now.

“We’ve heard publishers’ feedback on our proposed ads and business terms update,” Stone said. “We care deeply about our partners and do not plan to make policy changes without a full consultation with industry groups and leaders. We’re committed to getting this right, and as we expect this consultation could extend beyond November 20, in which case we’d delay implementation of the updated terms.”

Meta has insisted that it would never use its terms of service to skirt robots.txt. In an update to business page owners, the company offered a “reminder” that “our crawling practices remain unchanged, and content owners can continue to signal their crawling preferences using robots.txt.”

But the company still faced criticism after promising to pull the terms and consult with publishers to get the policy right.

As regulatory and court scrutiny intensified, Nieman Lab noted that Meta took steps to try to assuage publishers’ concerns but “has not included any specific language about respecting robots.txt preferences in its revised terms of service.”

An infant in Wisconsin has died of measles, the first measles death in the state since 1990, officials reported Thursday. The death comes amid an outbreak in the Badger State that has stricken over 430 people across 13 counties this year.

“Any death of a child is a tragedy, and even more so when it is related to a vaccine-preventable disease,” Wisconsin Health Secretary Kirsten Johnson said in a statement. “Measles can cause serious illness and death, and it is important for all Wisconsinites to evaluate their risk and take steps to protect themselves and their family.”

The baby’s death is the third measles fatality nationally just this week, following the death of a New York child reported on Monday and another death in Pennsylvania reported on Wednesday. The Pennsylvania death was of an unvaccinated person in Jefferson County; the state health department did not provide additional details, such as the age of the person. The state’s explosive, ongoing measles outbreak has led to 1,467 cases so far this year.

In all, there have been nine reported measles deaths this year in the US, eight of which have been confirmed by state officials. The deaths include the baby’s death in Wisconsin, the child’s death in New York, and seven deaths (six confirmed) in Pennsylvania, including two babies, a healthy teen, and several adults with underlying health conditions.

National context

As of November 5, the Centers for Disease Control and Prevention is reporting 5,182 measles cases in the US this year across 47 states and jurisdictions. The CDC is reporting four measles deaths, though it’s unclear which of the reported deaths the CDC has counted.

The national case and death tallies are the highest recorded since 1991, though the reported cases likely represent a significant undercount of actual cases. As Wisconsin health officials noted in the announcement of the baby’s death, “Not everyone who becomes ill seeks medical care or gets tested, which means cases may go unreported.” This is particularly true given that large outbreaks, including the one ongoing in Pennsylvania, have struck in tight-knit religious communities that are less likely to seek conventional medical care.

In 2000, the US declared measles eliminated after a year without continuous domestic transmission, thanks to high vaccination rates. The declaration was a public health triumph over the virus, which is among the most contagious on Earth. Prior to vaccination, measles sickened 3 to 4 million people per year, sending nearly 50,000 to the hospital, causing 1,000 cases of brain swelling, and killing 400 to 500 people.

Amid slipping vaccination rates nationwide and the spread of anti-vaccine misinformation and rhetoric, the virus has come roaring back, and the country’s elimination status is likely to be stripped. Next week (November 10 to 13), a committee of infectious disease experts assembled by the Pan American Health Organization—a regional office for the World Health Organization—will review data from 2025 and early 2026 to assess if measles has reestablished in the US.

Politicized public health

Under anti-vaccine Health Secretary Robert F. Kennedy Jr., much about the vaccine-preventable outbreaks has become political and contentious, including the definition and confirmation of measles-associated deaths. Amid the first deaths reported in Pennsylvania, Kennedy falsely suggested state officials “fabricated” the deaths and reportedly ordered the CDC to delete them from its national data.

The CDC then changed the way it counts measles deaths, abandoning a long-standing practice of reporting state-confirmed deaths. Now, the CDC relies on federal death records, which take much longer to process and are not intended to provide real-time data during active outbreaks.

The seventh, unconfirmed Pennsylvania death was reportedly of a 39-year-old woman in Franklin County; it was announced on October 12 by the county’s Coroner’s office. At the time, the coroner explicitly said the death met the state’s definition of a measles-associated death. Under the state’s criteria, a death is considered measles-associated if the person dies within 30 days of testing positive and shows clinical signs of infection, with no other significant cause for the death, such as a car crash.

Franklin County Coroner Jordan Conner listed the death as “a probable acute cardiac event due to possible complications of the measles” and reported that the woman had recently been diagnosed with and treated for measles.

In an email exchange with Ars Technica on Thursday, the state health department said it “has not found any Franklin County cases” that meet the state’s criteria but that “the investigation has not yet been closed.” The Franklin County coroner did not respond to a request for comment on the discrepancy. The Philadelphia Inquirer has taken legal action to obtain more information about the woman’s death, including her name, which was withheld by the coroner and sealed by a county judge.

This post was updated to correct the name of the Franklin County Coroner. 

Yesterday, when reporting on emerging efforts to pipe the console version of GTA VI through a PC’s DLSS-powered Nvidia GPU, we mentioned in passing that PC players could instead just wait for the AI-assisted decompilation port that will no doubt be coming in short order. Now, a former Rockstar tech director says he thinks that exact threat could cause the developer to work faster on the official PC version of its soon-to-release crime/lifestyle simulator.

Obbe Vermeij hasn’t worked at Rockstar since 2009, but in a new interview with Rock Paper Shotgun, he recalled how artists for the PC version of the original top-down Grand Theft Auto were incensed when the console version squashed the game’s 256-color textures down to 16 colors for the PS1 release.

“So it always has been that sort of awkward relationship between PC and the console,” Vermeij said. “So in those days, it was like PC first and then console second, even though that didn’t really make much sense for the sales, because the PlayStation versions sold way better.”

Those sales caused Rockstar to pivot hard toward the console side of the business, with oft-delayed PC ports of its major games only hitting many months or even many years after their console counterparts. It’s a trend that Rockstar has stuck with even as many formerly console-focused game publishers have started treating the PC as an equally relevant platform and as the line between PCs and consoles itself has gotten much fuzzier in recent years.

When it comes to major 3D releases, Rockstar clearly prioritizes console launches over PC ports.

Credit: Kyle Orland

Time is getting short

The rise of AI-powered reverse-engineering tools—which can quickly turn a console game binary into easy-to-port raw code with incredible brute-force efficiency and minimal human effort—might change that kind of “PC can wait” calculus for Rockstar. “I think I would be concerned, if I was Rockstar, that somebody would just take the PS5 version and get AI to change it to PC,” Vermeij told Rock Paper Shotgun.

To be sure, decompiling a game as complex as Grand Theft Auto VI wouldn’t be as simple as, say, the previous human-led decompilations of early 3D GTA games. But even if an AI coding harness has some difficulty capturing all the details of GTA VI‘s advanced weather systems and scripted sweat simulations, that doesn’t mean it’s impossible.

Vermeij speculated that reverse-engineering the console version of GTA VI could cost “tens of thousands of dollars in tokens,” and a resulting port probably “won’t be as good as [Rockstar’s] effort” in the end. That said, Vermeij said it would probably be good enough to satisfy some potential PC players and that he “wouldn’t be surprised if there was an effort to get [the releases] a little bit closer together” to get ahead of that.

“What was the case before is that there’s so much effort in developing the game, that it is not possible to rip it off, and that’s completely changed now,” Vermeij said. “So, yeah, we’re going to have to see in the next couple of years how that’s going to pan out.”

While there hasn’t been an official timing announcement for the PC port of Grand Theft Auto VI, reports suggest Rockstar is currently aiming for the first quarter of 2027.

Apple’s new MacBook Pro is expensive as hell.

Let’s get that out there right away, because for a lot of people, it will be the only thing that matters. As with so many things in life, though, you get what you pay for. There’s no questioning the quality of the laptop’s display or most of its other hardware components.

But should you buy it? Well, that’s a complicated question in 2026.

We’re in a time of great uncertainty when it comes to buying a computer. Prices are at levels we’ve not seen in recent history, and they may well keep climbing. Meanwhile, users paying close attention to the early developments in both local and agentic artificial intelligence are asking an important question: “What will computing even look like in a few years, and what does that mean for the computer I should buy today?”

Apple’s new MacBook Pro sits at the intersection of all of this uncertainty. In multiple ways, it’s also straddling different worlds.

It arrives amid a changing of the guard in Apple’s leadership, potentially raising questions about the future direction of Apple’s ecosystem. It’s a mass-market consumer product, but configuring it with enough RAM and storage to make it feel future-proof pushes its price into territory mostly only well-funded businesses can afford. It’s firmly rooted in the computing paradigms of the past 20 years, but it shows early signs of flirtation with new AI workflows. It’s a conventional Apple laptop in most ways, but it also has a touchscreen—something that would never have happened in a prior Apple era.

It’s an outstanding, high-quality product, but whether it’s something you should buy depends on you—what you’re doing right now, what you’ll be doing in the future, and what you value.

Specifications and performance

For most of the specs we look at, Apple isn’t claiming any difference here. The new MacBook Pro promises 24 hours of battery life (though you’d have to use it in a conservative way to achieve that), it has the same ports (three Thunderbolt, one HDMI, one SDXC, and MagSafe), and it has the N1 chip for Wi-Fi 7 and Bluetooth 6 connectivity.

And like its immediate predecessor, it offers the M5 Pro and M5 Max chips as options. Running synthetic benchmarks, we found no meaningful difference at all compared to when we ran the same tests on the previous model, so we won’t spend time discussing their performance today. We’ve covered both thoroughly in prior reviews, and you can read up there if you’d like.

The chips are more than fast enough for normal workflows for the foreseeable future, including most 3D work. AI inference is the main area where they may be less future-proof. To be clear, they’re strong enough for now, but the field is evolving rapidly, and there’s reason to expect significant improvements in future chips on this front.

According to Bloomberg, Apple is skipping Pro and Max versions of the M6 chip that was introduced earlier this year to focus its attention on a significant rework for the M7 due in 2027. “The M7 line is designed primarily around major advancements to on-device AI processing,” Bloomberg’s Mark Gurman wrote. The line would emphasize memory bandwidth, which matters for fast, efficient AI inference.

Even without Bloomberg’s report, I’d expect Apple to prioritize those areas in its next high-end chip refresh. That’s partly because there’s another tea leaf to read here: Apple made just such an update with the A20 Pro, which debuted in its latest iPhones several weeks ago. That update brought a substantially improved Neural Engine and a significant increase in memory bandwidth. We also found that the A20 Pro delivered improved 3D graphics performance in some cases.

Similar improvements could come to Macs, meaning it might be worth waiting another year if you’re focused on AI workflows or even 3D applications—especially given the high cost of these machines. That’s part of what makes this such an uncertain time to buy a laptop. We’re in a moment when these companies are revisiting the architecture of these chips because of these new workflows, and we don’t know what the end result is yet.

A monster of a display

Here’s one thing that’s not at all uncertain: The display on the MacBook Pro is the best I have ever used in a laptop. If the display is your primary concern, this is a big moment.

Apple calls it a “tandem OLED display,” meaning a panel with two emissive layers. This approach helps achieve a higher peak brightness, but perhaps more importantly, it can reduce the risk of burn-in. Most of the time, the display won’t need both layers running at full power. Instead, they can share the load and run at a reduced current, helping extend the panel’s lifespan.

The new display can achieve up to 2,000 nits of peak brightness on HDR highlights, but it’s rated for 1,000 nits of sustained brightness. In general, I like the way Apple handles HDR for this at the OS level, too.

This photo doesn’t do the screen justice, but at least you can see the proportions and design here. (It’s the 16-inch model.)

Credit: Samuel Aoxn

The brightness matters for HDR content, sure, but what really makes OLED stand out is per-pixel illumination—meaning each pixel can turn on or off independently. That allows for absolute black levels and very strong contrast. The MiniLED display in the previous MacBook Pro offered good contrast for an LCD display, but OLED has a fundamental advantage here, and it shows.

As you’d expect, the screen offers native support for both P3 and Adobe RGB color gamuts. It also has a variable refresh rate from 10 to 120 Hz, just like many of Apple’s other products. As before, the MacBook Pro comes in roughly 14-inch (3,024×1,986 pixels) and 16-inch (3,472×2,226 pixels) variants. There’s also a nano-texture glass option to fight glare, though that will set you back another $150.

It’s one of the most impressive displays I’ve used in any device, and it’s certainly the best I’ve ever used on a laptop. The colors are vibrant, the contrast is outstanding, the blacks are inky, and it can get very bright when it needs to—not just with highlights but full-screen brightness as well. And on the whole, it’s extremely precise. Uniformity appears basically flawless. It’s what you’d expect from an ultra-high-end TV or reference monitor, not a laptop or computer screen, and it’s genuinely striking.

We’ll see about longevity, but OLED has come a long way in recent years, and my first impressions are that Apple has made all the right choices here.

I find myself trying to find something negative to say about the MacBook Pro’s screen to avoid sounding too effusive, but I can’t think of anything substantive. The best I can do is complain that Apple’s proprietary “PROLED XDR” label is silly and eye-roll-inducing.

Branding aside, the screen is the main reason to pay extra for this model.

Design

With its lid closed, the new MacBook Pro looks identical to its predecessor. The dimensions are largely the same. It’s slightly thinner, but you’d have to check the spec sheet or handle them side by side to notice.

Pick it up, though, and you’ll find that it’s noticeably lighter than the last model, by about half a pound. It’s not in a totally different weight class, but you can notice the change, and it’s a nice improvement. The reduction is partly due to the display technology switch; it’s easier to make a light OLED screen than a light MiniLED one.

Open the MacBook Pro and turn it on, and you’ll see one of a few other notable differences, depending on whether you’re behind or in front of it. From behind, you’ll see the Apple logo on the lid now lights up, as it did on MacBooks of yore. It’s arguably a little tacky, but it’s also neat and nostalgic. I don’t hate it.

The glow returns!

Credit: Samuel Axon

From the front, you might notice that the already barely there bezels are even more barely there, though that’s subtle. More likely, you’ll see that the top-of-screen camera notch has been replaced by a cutout just like the Dynamic Island Apple introduced on the iPhone.

A comparatively sensible camera cutout

When I reviewed the first iPhone with a Dynamic Island, I was largely positive about it. No, it’s not ideal to have a cutout on the screen, but Apple has handled it as intelligently as possible, and I would even go so far as to say it has turned what should be an albatross into an attractive feature.

Thanks to the OLED display, it’s possible to make any of the pixels around the camera cutout just as black as it is. Apple has designed live activities, the volume bar, and more to appear as if they are expanding out of and around, and contracting back into, this little dot. It remains ingenious.

When the Dynamic Island first arrived on the iPhone, it helped unify several previously disparate approaches to displaying notifications in iOS, making it especially useful. This time, it doesn’t solve as many problems, but it’s an improvement over the dead weight of the notch. It also helps that the MacBook Pro’s much larger screen means the cutout takes up a small share of the screen’s real estate.

The Dynamic Island on the MacBook Pro.

Samuel Axon

The Dynamic Island on the MacBook Pro. Samuel Axon

Full-screen apps cut off just below the Dynamic Island by default, just like with the notch before.

Samuel Axon

Full-screen apps cut off just below the Dynamic Island by default, just like with the notch before. Samuel Axon

The Dynamic Island on the MacBook Pro. Samuel Axon

Full-screen apps cut off just below the Dynamic Island by default, just like with the notch before. Samuel Axon

The Dynamic Island works the same with full-screen apps as the notch did. By default, the top of the viewport ends just below the Dynamic Island, so it just looks like there’s a taller top bezel, and it doesn’t cut into the application interface. The menu bar, clock, and other system elements still occupy the space above the application (set against a totally black background) when you move the cursor up there.

I tested a few fullscreen games with it, like World of Warcraft and Cyberpunk 2077. Some behaved by default like the default fullscreen app behavior I just described, but others tried to occupy the whole thing, making the cutout visible. Vitally, every video app I tried avoided the cutout in fullscreen.

This isn’t something to upgrade for, of course. But on the whole, I think it’s an improvement over what we had before.

The most touchable MacBook Pro ever

Apple has finally added a touchscreen to the MacBook Pro. I say “finally” because the possibility has floated around for about two decades. Steve Jobs was famously opposed to the idea; there’s even video of him explaining why he thought it would make for a poor user experience.

A couple of years ago, reliable reports emerged that Apple was moving forward with the idea regardless. When I first heard about it, I imagined what could be a really exciting implementation: a convertible laptop that would let you use it in a tablet-like mode, with Apple Pencil support for drawing and writing.

As a game developer, I imagined quickly sketching 2D game maps with an Apple Pencil, using a local AI model to convert them into 3D geometry, and then switching to a laptop configuration to start testing and coding. I also imagined using it like an iPad to more comfortably watch movies on a flight.

That’s not what we got, though. The laptop has the same form factor as the previous MacBook Pro model—no tablet mode or Pencil support. Instead, it just lets you reach up from the keyboard and trackpad to touch the screen. That feels like a missed opportunity.

Some creative apps have implemented touch features that might justify the touchscreen functionality if you use them a lot. For example, you can play the on-screen virtual piano keys on Logic Pro with multiple fingers, which could be nicer than clicking on them. I imagine many people already have workflows built around other approaches, though.

If you were specifically waiting for a touchscreen on your MacBook Pro, this seems like as good an implementation as you could hope for without turning the laptop into a convertible. Apple has put real care into it. For example, it has re-engineered the hinge so the screen is not particularly wobbly when you touch it. It’s fast, accurate, and responsive. It supports multi-touch gestures like pinch-to-zoom or swiping with three fingers to switch desktop spaces.

There’s one use case I find compelling as a developer. Testing iPhone and iPad apps in the Simulator has always been a minor hassle because of the difference in input methods. You either have to approximate only a subset of the interactions available to users with a trackpad and mouse or connect an iPhone or iPad to test them that way. A touchscreen could provide a more elegant way to test such apps in Xcode, Unity, or whatever else you might use.

Unfortunately, some of the UI elements in macOS are too small for my large fingers, so I have to tap very carefully. I also keep forgetting the touchscreen is even there for hours at a time and go back to my normal habits, partly because there’s nothing you can do at the OS level that you can’t do just as easily with trackpad gestures. And as Jobs predicted all those years ago, my arm got tired while I was using it.

For me, it feels faster, easier, more comfortable, and more natural to use the keyboard and trackpad 100 percent of the time, but I admittedly have decades of muscle memory and habits built around them. There’s apparently a generation coming up with the same kind of familiarity with touch interfaces, so this feature may appeal to them. Plus, touchscreen laptops have been a thing on Windows forever, and some people seem to want to buy them, though I never quite got why without an added tablet configuration.

It’s unlikely to account for a big portion of the device’s cost compared with the rest of the OLED panel, memory, storage, and so on. So, my own bafflement aside, I’m not unhappy it’s here. I imagine it will serve those who want it well enough.

The same excellent workstation—if you can afford it

As noted earlier, whether this laptop is worthy of a recommendation depends on you and how you use your machine. That’s always been the case, but it seems more so here.

If you’re a creative professional or a display enthusiast who wants the best laptop screen you’ve likely ever seen, you won’t be disappointed if you shell out for it. And hey, maybe you can show me your workflow and prove me wrong about the touchscreen!

If you’re just looking for a great laptop with strong performance for everyday use, consider the non-OLED, non-touchscreen model that’s still in Apple’s lineup—or even a MacBook Air, which compromises relatively little for most users. And there is, of course, the entire world of Windows and Linux laptops.

If you’re a developer—especially one working at the frontier of AI workflows—well, that’s where things get tricky. Will waiting be worthwhile? So much about how those workflows will evolve and what they’ll demand from your hardware is still so up in the air that I can’t make a confident recommendation. (For what it’s worth, I fit best in this category, and I’m still holding onto my M1 Max machine, though my resolve is wavering.)

All that said, in most ways, this is still the same MacBook Pro we’ve reviewed again and again for the past few years. Touch adds a feature aimed at a particular niche, or perhaps it’s there to roll out the welcome mat for a newer generation. The display is a more substantial update and makes a compelling case for upgrading, especially if you’re coming from an older, pre-MiniLED laptop.

But this is an expensive machine. The configuration I reviewed here (16-inch, M5 Max, 128 GB of RAM, 2 TB SSD) retails for a whopping $7,399, though you could at least shave $1,600 off that by going to a more typical-for-most-people 64 GB. (The cheapest configuration starts at $2,899.)

For the vast majority of buyers, the pre-OLED version that’s still in Apple’s lineup or another MacBook altogether will offer more bang for your buck. If you’re the kind of person who wants OLED or a touchscreen, though, you probably know that already. Based on my time with the new MacBook Pro, both features deliver on their promises, even if the latter isn’t as ambitious as some of us had hoped.

The good

• Outstanding display that offers notably better and more accurate image quality while modestly reducing the weight of the device
• The M5 Pro and M5 Max are still plenty fast for the majority of users
• Strong build quality, solid port selection, and everything else we had positive to say about the last MacBook Pro, too

The bad

• For certain workflows, the M5 Pro and M5 Max are already a few months old in an environment where things are moving very quickly
• The RAM and SSD price crunch is real, and these prices are a formidable barrier for most people

Top Republican lawmakers have asked Jensen Huang to appear before Congress to answer questions after it was reported that US intelligence officials believe Nvidia’s chief executive has been targeted by a Chinese spying operation.

The chairs of the House committees on China, foreign affairs, and intelligence wrote to Huang this week asking him to appear before a joint closed-door briefing for their committees on November 18.

“Recent reports raise significant national security concerns regarding the People’s Republic of China’s access to sensitive US technology and information,” representatives John Moolenaar, Brian Mast, and Rick Crawford wrote.

“The issue of PRC infiltration and access to cutting-edge US technology is core to the investigative work of our three committees this Congress. It is vital that Congress determines the extent to which US national security has been placed at risk,” the lawmakers added.

The New York Times last week wrote that US intelligence officials believed Huang had been the target of a Chinese spy operation. The concerns center on a Chinese woman named Alice Huang, an executive at Megaspeed, a Singapore-based cloud company that has bought billions of dollars of Nvidia’s chips.

Nvidia has rejected the claims and on Thursday would not say if Huang would appear before Congress. “Meetings with customers are part of a CEO’s job. Jensen talks with thousands of customer executives every year,” the company said.

“Our regional cloud customers receive ordinary business information—not sensitive Nvidia trade secrets—and zero information relevant to national security,” it added.

The lawmakers said they wanted Huang to discuss the security protocols Nvidia had in place to protect against “foreign adversary espionage.” The company said its chief executive had an “outstanding security team” that travels with him.

The New York Times said Huang had not received a “defensive briefing,” which the intelligence community usually, but not always, provides potential espionage targets to warn them about the activity.

Nvidia said it was confident that the US government would have let the company know if it thought a person was a threat to US industry and that it often receives updates from Washington about people of concern.

The US has previously scrutinized Megaspeed, founded in 2023 as a spin-off from a Chinese gaming company, for potential chip smuggling, although no charges followed.

Nvidia launched its own unannounced on-site inspections at Megaspeed in Singapore when it became aware of the US concerns, but the internal probe found no evidence of its chips being diverted into China, said people familiar with the matter.

The company’s supply chain has come under increasing scrutiny as US export controls have prompted increased smuggling of its chips. Earlier this year Wally Liaw, a co-founder of Supermicro, one of Nvidia’s key supply chain partners, was arrested in the US and charged with smuggling Nvidia chips into China. He denies any wrongdoing and has pleaded not guilty.

© 2026 The Financial Times Ltd. All rights reserved. Not to be redistributed, copied, or modified in any way.

Elon’s Edsel is no more. Last night Tesla CEO Elon Musk revealed on social media that his car company is calling time on its failed attempt to wrest control of the American pickup truck market. The Cybertruck will cease production in the middle of 2027, liberating that space in Tesla’s cavernous factory for other uses—more Cybercabs, most likely.

Musk first revealed the Cybertruck in late 2019, back when Tesla was still trading at a theoretically understandable price-to-earning ratio. The pickup’s unusual angular shape owed more to low-res video game models than any concessions to practicality. At the time Musk made fanciful claims that the truck would use a stainless steel monocoque construction, telling his audience that “the body and bed don’t do anything useful” on a normal truck. While some race cars use carbon-fiber monocoques for strength and lightness, making a pickup truck with a stainless steel monocoque chassis would only make sense if someone had an awful lot of stainless steel lying around and needed something to do with it all.

The truck was supposed to go on sale for less than $40,000, and Tesla made its usual stratospheric sales predictions that (had they been accurate) the bare metal EVs would be as ubiquitous as the Model 3 and Y.

But development of the Cybertruck was more troubled than usual for Tesla, which struggled heavily to bring both the Model 3 and Y into mass production. Musk demanded “Lego-like precision” from the workers assembling the trucks, but achieving the 10-micron tolerances required by the Danish toy company was not evidently successful, based on what we saw when we experienced our first Cybertruck in reality.

That was in 2023, when Cybertruck headlines were mostly about recalls and stop sales. By then, the truck used a conventional unibody chassis—conventional for Tesla, not for full-size pickup trucks. Toward the end of that year, Musk told his audience that “we dug our grave” with the truck, perhaps the first public acknowledgment that annual sales of a quarter of a million Cybertrucks probably wasn’t going to happen after all.

Sales never took off, and the divisive shape of the Cybertruck became an all-too visible reminder of the CEO’s role in reshaping the federal government to his liking. Even selling thousands of Cybertrucks to SpaceX couldn’t save the pickup. Now, with America having vociferously rejected Musk’s vision of governance, those wishing to drive something that reminds everyone around them of that vision will only have a few months more to purchase one.

Welcome to Edition 9.18 of the Rocket Report! Blue Origin packed a lot of news into one afternoon Thursday with the reveal of a new kick stage for the New Glenn rocket, a new rocket engine design, and the minor and widely expected delay for New Glenn’s return-to-flight mission from December until January. The reconstruction of Blue Origin’s wrecked launch pad at Cape Canaveral, Florida, continues apace after a catastrophic explosion nearly six months ago. There’s a lot of pent-up demand from NASA, Amazon, and the Space Force for Blue Origin to get New Glenn back in action.

As always, we welcome reader submissions, and if you don’t want to miss an issue, please subscribe using the box below (the form will not appear on AMP-enabled versions of the site). Each report will include information on small-, medium-, and heavy-lift rockets as well as a quick look ahead at the next three launches on the calendar.

SpaceX vets work on water-fed propulsion. Three brothers who worked at SpaceX in the Falcon 9 era are now building satellites that run on water, a potentially harvestable resource on the Moon and found all over the Solar System, Ars reports. In late 2022, the three Carlisle brothers founded Argo Space to build infrastructure in space. Their vision was not just in-space transportation but also removing the need to use propellants from Earth to move around the heavens. Argo Space’s satellites will use water not only as a propellant, but also as a coolant and radiation shield.

How does this work? … Argo Space flew its first mission in July, but the spacecraft failed after one of its solar arrays didn’t deploy, and there wasn’t enough time to test the propulsion system. Two more missions are on Argo Space’s schedule for 2027 to demonstrate the company’s water-based thrusters. The propulsion system is basically a small rocket that uses hydrogen and oxygen as propellants. However, instead of launching with separate tanks of liquid hydrogen and liquid oxygen, the satellite carries a reservoir of liquid water. In space, the vehicle uses solar power to split water into hydrogen and oxygen gases and burns them to generate thrust.

JAXA backs small satellite launcher. Japanese launcher startup Interstellar Technologies has secured the backing of the Japan Aerospace Exploration Agency (JAXA) for its Zero rocket program, Aviation Week & Space Technology reports. The company said JAXA selected it as part of its Acceleration of Commercial Rocket Launch Demonstration program under the Space Strategic Fund. The government-funded program supports development and flight demonstrations of private launch companies with grants of up to $76 million per company, with the goal of fostering a growing commercial launch industry in Japan. Interstellar previously raised more than $250 million, with roughly equivalent funds from government and private capital.

Zeroing in … Interstellar is using the money to complete development of its Zero rocket, a two-stage launcher designed to carry up to 1 metric ton (2,200 pounds) of payload to low-Earth orbit. “We have achieved critical technical milestones, including Japan’s first successful firing test of an orbital-class methane engine,” said Satoshi Nakayama, director of the board and president of Interstellar. “We have now reached the stage where we must swiftly turn these proven development results into operational commercial services.” The most recent update from Interstellar indicated a first launch of Zero in 2027.

Launch demand outpacing supply. A new report from analytics firm Quilty Space confirms that, outside of China, the world faces a significant shortage of rockets over the next five years, Ars reports. The firm estimates that between now and 2030, there will be a shortfall of about 150 launches compared to demand from satellite operators and civil space agencies. If demand from the US military is added in—and SpaceX does not extend the Falcon 9 and Falcon Heavy program to accommodate increased national security needs—the shortage could be doubled.

Worse than we thought … “What we wanted to get at is to understand how severe this launch crisis is and when will it be resolved,” said Caleb Henry, director of research at Quilty Space. “What we learned is that the launch crisis is more severe than we thought and that it will not go away this decade for sure, and it might not abate until sometime late in the 2030s.” Why is this happening? Quilty puts the blame on several factors: the war in Ukraine taking Russian rockets off the Western market; Amazon Leo demand; developmental delays in a new generation of rockets, from New Glenn and Vulcan to Neutron, Terran R, and Nova; and more recently, the retirement of the Falcon 9 as a rocket for commercial service.

China continues launching mega-constellation. Another 18 satellites for China’s SpaceSail Internet network launched Tuesday (Monday, US time) from a spaceport in the country’s northwestern desert. The mission marked the second launch—and first operational flight—of the Kinetica 2 rocket developed by a Chinese startup known as CAS Space. The SpaceSail constellation, also known as Qianfan, now consists of some 275 flat-panel satellites from a series of launches that began in August 2024. The network’s first phase will number 1,296 satellites. 

A mini-Falcon Heavy … The Kinetica 2 is an interesting launch vehicle, with three side-by-side boosters arranged in a manner similar to SpaceX’s Falcon Heavy. But the Kinetica 2 is a much smaller rocket, with three kerosene-fueled engines on each booster instead of nine. Its lift performance to low-Earth orbit is about 12 metric tons (26,000 pounds). CAS Space plans to eventually recover and reuse the Kinetica 2 boosters.

India launches strategic navigation satellite. A Geosynchronous Satellite Launch Vehicle (GSLV) rocket launched October 30 with a satellite to restore India’s regional navigation system, NDTV reports. The NVS-03 satellite is the third spacecraft in India’s second-generation navigation network. Some of the fleet’s older satellites suffered from atomic clock failures, rendering them partially or totally unusable. NVS-03 carries multiple clocks of different designs in an attempt to overcome the system’s reliability problems.

A “naughty boy” … India’s medium-lift GSLV has now launched 20 times since debuting in 2001. The rocket doesn’t fly as often as India’s smaller launcher, the PSLV, and hasn’t always performed well when it has. Six of the 20 flights ended in total or partial failures, but most of the mishaps occurred early in the rocket’s history. GSLV’s track record earned it the title “naughty boy” from Indian space officials. The most recent launch was a different story. The head of India’s space agency, V. Narayanan, declared the GSLV’s recent flight was the “most precise launch of any rocket in the world.” We’re not sure what metric Narayanan used to make that claim, but the launch was, indeed, a success.

China reveals roadmap to lunar landing. Chinese space officials confirmed that the Asian nation plans to land its astronauts on the Moon within three years, Ars reports. At the Seventh Manned Spaceflight Conference in Beijing, officials, including Wang Xiang, chief designer for the lunar exploration program, laid out the country’s most publicly detailed plans yet for a series of missions leading up to a crewed lunar landing “before 2030.” China’s lunar landings will involve a lunar-orbit rendezvous similar to what NASA did in the 1960s and early 1970s with the Apollo program. However, lacking a rocket as powerful as the Saturn V vehicle, the crew and lunar lander will launch separately to the Moon.

These are the rockets … China’s Long March 10 rocket family will be the workhorse for the country’s initial crew missions to the Moon. The next big test will be the demonstration flight of a Long March 10A rocket, which flies with a single core booster, and China’s new Mengzhou Moon ship to low-Earth orbit in early 2027. The larger Long March 10, with three boosters connected together, will launch China’s Lanyue lunar lander on an uncrewed test flight to the Moon in 2028. The next year, a follow-up mission will involve the concurrent launch of two Long March 10s, one with an uncrewed Mengzhou spacecraft and the other with a Lanyue lander that will dock together in lunar orbit. Then, before 2030, another dual launch of Long March 10s will occur, this time with crew on the Mengzhou spacecraft, leading to the crewed landing mission.

Blue Origin’s big reveals. Blue Origin announced two significant upgrades to its New Glenn rocket on Thursday, reducing the cost of manufacturing the vehicle and expanding its capacity for payload delivery to geostationary orbit and the Moon, Ars reports. With these upgrades, the company aims to eventually position New Glenn to launch more than 1,000 times a year. One of the announcements involved a new engine, called Northstar, that will eventually replace the BE-4 engine that powers the first stage of New Glenn. The other piece of news was that Blue Origin is developing a hydrogen-oxygen third stage for the rocket to give it an extra kick for high-energy missions.

Blue’s Northstar … The Northstar engine is a simplified but more powerful version of the BE-4 engine. It has half the number of parts and 30 percent less mass, and about 65 percent can be 3D-printed, all while providing a thrust increase over the BE-4. Blue Origin believes it can manufacture the engine at 25 percent the cost of a BE-4 engine. The kick stage is somewhat of an engineering marvel in itself. The stage will use a lightweight composite fuel tank, but officials were not sure the structure could contain leak-prone hydrogen at high pressure. The company recently had a breakthrough at Launch Site One in Texas, when it completed its first composite liquid hydrogen tanking test, giving engineers confidence to move forward with the design.

New Glenn won’t return to flight this year. Blue Origin now plans to resume launches of its New Glenn rocket in January, delaying the much-anticipated flight of the vehicle after a destructive launch pad explosion in May, Bloomberg reports. The Jeff Bezos-backed company had been driving to launch New Glenn again before the end of the year in its quest to challenge SpaceX. But Blue Origin Chief Executive Officer Dave Limp told employees on Wednesday that they are now targeting a flight early in 2027.

What’s a month? … The slight delay is unsurprising, and a January launch would still put Blue Origin well head of initial fears of how long it would take to rebuild its launch pad in Florida. New Glenn has a busy launch manifest after the return-to-flight mission, including a crucial launch in the spring of Blue Origin’s Blue Moon Mark 1 lander, an uncrewed precursor to the company’s future human-rated lander for NASA’s Artemis missions.

Artemis II tore through the atmosphere. The launch of NASA’s Artemis II mission in April caused significant disturbances in Earth’s ionosphere that could have triggered temporary, localized errors in GPS positioning, a new study has found, Space.com reports. The study by a team of researchers from National Cheng Kung University in Taiwan analyzed data from high-precision networks of fixed, ground-based GPS receivers. The researchers used the GPS measurements to reconstruct how the passage of the Artemis II mission’s Space Launch System rocket changed the concentrations of free electrons in the upper atmosphere.

Disturbing the peace … The measurements found that the ionospheric disturbances resulting from the Artemis II launch were more significant than any caused by other rocket launches to date. Curiously, they were up to 10 times more intense than those observed following the test launches of SpaceX’s Starship, which is twice as powerful as the SLS based on liftoff thrust.

Next three launches

November 6: Ceres 2 | Unknown Payload | Jiuquan Satellite Launch Center, China | 23:35 UTC

November 7: Falcon 9 | USSF-432 | Vandenberg Space Force Base, California | 00:46 UTC

November 8: Long March 12 | SatNet LEO Group | Wenchang Space Launch Site, China | 20:25 UTC

KENNEDY SPACE CENTER, Fla.—Blue Origin announced two significant upgrades to its New Glenn rocket on Thursday, reducing the cost of manufacturing the vehicle and expanding its capacity for payload delivery to geostationary orbit and the Moon.

With these upgrades, the company aims to eventually position New Glenn to launch more than 1,000 times a year.

“There has never been higher demand for launch, ever,” Dave Limp, chief executive officer of Blue Origin, said during the UP.Summit conference at the Shuttle Landing Facility in Florida.

Blue Origin has already begun test firing a new engine, called Northstar, that will eventually replace the BE-4 engine that powers the first stage of New Glenn. Separately, the company also said it was developing a hydrogen-oxygen third stage for the rocket, GS3, to give it an extra kick.

A new Northstar

The Northstar engine is a simplified but more powerful version of the BE-4 engine, Limp said. It has half the number of parts, 30 percent less mass, and about 65 percent can be 3D-printed.

Blue Origin believes it can manufacture the engine at 25 percent of the cost of a BE-4 engine, and eventually reach the point where each engine can be assembled in less than a day. The company is deep into development and testing of Northstar, having already begun a test-firing campaign.

“To get access to space at a lower cost, we have to look at every part of the vehicle,” Limp said. “This is going to be a game-changer.”

The BE-4 engine powering New Glenn today has a thrust of 640,000 pounds, and Limp said Northstar will have a starting thrust of 715,000 pounds. Like the BE-4 engine, Northstar burns methane and liquid oxygen using an oxygen-rich staged combustion cycle and a single-shaft turbopump architecture.

Limp said the company will continue to manufacture the existing BE-4 engine for United Launch Alliance, which uses two of the engines to power its Vulcan rocket.

GS3 provides a nice kick

Blue Origin is developing a third stage for New Glenn called GS3 (creatively named Glenn Stage 3) to offer direct injection to geostationary orbit and Earth-escape trajectories. The third stage fits within an interstage, preserving the full fairing volume for customer payloads.

The stage can fly on both the smaller 7×2 variant (as in seven first stage engines and two second stage engines) of New Glenn as well as the larger 9×4 variant. According to Blue Origin, a GS3-equipped New Glenn 7×2 can carry 8 metric tons directly to geostationary orbit and 10 metric tons to translunar injection. Those figures increase to 14 tons and 20 tons, respectively, for the larger 9×4 variant.

As the rocket’s third stage, it needs to be light, Limp said. For this reason, the third stage is made mostly of composites, which are high-performing and lightweight. But there’s a problem with that.

The trouble with hydrogen

The stage will be powered by the company’s BE-7 engine, already used by its lunar landers. This engine is optimized for in-space operations and uses liquid hydrogen and oxygen propellants. But hydrogen is the smallest atom and therefore leaky. With composites, thermal cycling and mechanical strain can cause cracking, exacerbating losses of molecular hydrogen.

Limp said the company was not sure a large composite tank, 5.5-meter diameter in this case, could be constructed that would contain hydrogen at high pressures.

Blue Origin has successfully tested a 5.5-meter composite test tank in West Texas.

Credit: Blue Origin

The company recently had a breakthrough at Launch Site One in Texas, when it completed its first composite liquid hydrogen tanking test. During the test, the tank was pressurized to 100 percent of the maximum force expected during operation for 2.5 hours. It was even successfully tested at 150 percent of this maximum pressure for 10 minutes.

That gave the company confidence that a GS3 based on composite tanks was a concept that could go forward, and now the company is aiming for a debut in 2028 for a government customer.

“We have one customer signed up for cislunar space in 2028,” Limp said. “We are open for business.”

Devices running some configurations of Alpine Linux, and possibly other distributions, are affected by a decade-old vulnerability that allows attackers to log into devices with no password.

The vulnerability exists in the interaction between OpenSSH, the code library implementing SSH connections, and musl, a lightweight alternative to the libc library that, unlike most other Linux distributions, is used in Alpine. Further, OpenSSH using musl (pronounced muscle) must not use PAM (pluggable authentication modules), a framework that handles user authentication and account-access checks. Alpine is a lightweight, widely used Linux distribution. Other platforms may be affected if they use the same configurations.

One connection, works every time, needs no retries

There are two other conditions necessary for successful exploitation. First, a system must have at least one user account that’s accessible using a password that is hashed using the SHA256 or SHA512 functions. Second, the target system account must be configured to be passwordless, by way of specifying a “*” in the password field in /etc/shadow, the Linux file where password hashes are stored. Passwordless accounts are commonly used for accounts that use certificate-based authentication. This setting is common for certain types of accounts.

When all the conditions are met, the consequences are high.

“An attacker who can reach your SSH port can log in without knowing any password,” Alpine maintainers said Thursday. “It takes one connection, works every time, and needs no retries.”

When logging into a passwordless account, the attacker can gain access by inputting more than 256 characters in the password field.

Security firm Checkpoint said that the use of the SHA256 or SHA512 functions “matters because sshd borrows that account’s salt. Without it, the attack fails.”  Company researchers said their testing also showed that the account must be active, not just hold a hash. Checkpoint further reported that root accounts are safe under the default PermitRootLogin prohibit password setting, though reachable accounts with doas or sudo rights can still become root.

Despite the prerequisites, *sshole—the vulnerability name given by the person who discovered it—likely threatens large numbers of systems. The Docker Hub base image for Alpine has more than 1 billion downloads. Alpine has long been popular in routers, smart TVs, industrial controllers, and other types of embedded devices, many of which can’t be easily updated, if at all. More recently, Alpine has been adopted for use in desktops and servers. Alpine is also used in many container images.

According to security firm Detectify, Shodan lists around 22 million systems with SSH open on port 22. Only a fraction meet the conditions necessary for exploitation, but given the ease and speed of testing each one, locating vulnerable devices won’t require much effort or time.

It’s not entirely clear where the vulnerability resides. OpenSSH, which issued an update Thursday, said: “We consider the root cause of this bug to be outside sshd, but it is exploitable via sshd and can be avoided there.” Musl developer Rich Felker, meanwhile, wrote: “It is my position that this is not a bug in musl just an erroneous use by the caller.” Felker went on to say that musl developers have wanted to make changes to the mechanism “for a while now” and planned to do so in an upcoming release. Alpine has also issued an update.

It’s not yet known what other packages may be vulnerable. The person who discovered *sshole said OpenWRT is affected, but by default, that OS uses the Dropbear package, not OpenSSH, for SSH connections (though maintainers provide instructions for installing the latter). The discoverer also names the packages buildroot and yocto as being affected, but that isn’t yet confirmed.

Despite the uncertainty, anyone using a Linux system that uses both musl and OpenSSH should take time to immediately investigate if they’re affected. A good starting point is here. Both Alpine and OpenSSH patches should be installed on all affected systems. There is no known CVE designation for the vulnerability.

NASA has announced a new experimental aircraft, the X-70 Fractal, that will supposedly feature reconfigurable flight-control technology and a modular design for swapping in different payloads. But much remains unknown about the aircraft’s design and development timeline, even as NASA administrator Jared Isaacman said the X-plane could take flight within two years.

The mystery surrounding NASA’s newest X-plane was exemplified by the agency’s press release, including an apparently AI-generated image of the X-70, showing a dark silhouette vaguely resembling a military fighter jet obscured by smoke. The news was announced by Isaacman during the UP.Summit held at the agency’s Kennedy Space Center in Florida on November 4.

“The X-70 Fractal gives NASA a crewed, high-performance platform to test advanced flight-control, maneuverability, and other cutting-edge technologies in real flight,” Isaacman said. “Its modular design and ability to rapidly swap internal and external payloads will keep our teams flying, learning, and moving quickly from one experiment to the next.”

NASA’s press release described the X-70 as a crewed aircraft designed to cruise at Mach 0.75 while reaching a maximum altitude of up to 45,000 feet, and having an operational range beyond 1,726 miles (2,778 kilometers). The article did not mention a specific aircraft manufacturer that might be helping the agency develop the X-plane.

The X-70 would also have a takeoff weight of 10,000 pounds (4,536 kilograms) and a design capable of accommodating up to 2,000 pounds (907 kilograms) of experimental payloads both internally and externally. Ars has reached out to NASA for more details.

Return of the Blackbird?

Another X-plane is also expected to be announced at the UP.Summit on November 6, according to Aviation Week. That would coincide with a scheduled announcement from President Donald Trump at the event.

This comes after Isaacman previously appeared in front of another smoke-shrouded silhouette image with the tagline “Rebuilding our X-plane portfolio” during the All-In Summit in Los Angeles held from September 12–14, according to TWZ. That silhouette appeared strikingly similar to the SR-71 Blackbird that flew Cold War reconnaissance missions at supersonic speeds in the stratosphere.

Rumors have circulated about a possible revival of the SR-71 in one form or another, after former NASA employees told Aviation Week they received job offers to help restore an SR-71 to operational flight capability.

In September, the publication also reported on the removal of a NASA-owned SR-71 with the tail number 844 that had stood on display at NASA’s Armstrong Flight Research Center inside Edwards Air Force Base in California.

Quiet supersonic flight testing

Meanwhile, NASA recently commemorated the first flight anniversary of its supersonic X-59 Quesst, a needle-nosed aircraft designed by Lockheed Martin to reduce sonic booms to sonic thumps. Ars has previously looked at how this X-plane program is intended to support the design of future supersonic airliners that may fly overland while minimizing disturbances for buildings and people below.

Since first taking to the skies on October 28, 2025, the X-59 has completed over 40 flight tests, including both low-speed flights and supersonic test runs reaching Mach 1.53, according to NASA. Such testing has taken place at NASA’s Armstrong Flight Research Center inside Edwards Air Force Base in California.

The X-59 also recently completed its first cross-country journey from California to Florida with stopovers in Texas and Alabama, accompanied by an F-15 and an F/A-18 from the agency’s research fleet that served as chase planes. It sat on the ground as a static display during the NASA Max Power airshow at Kennedy Space Center in Florida from November 7–8.

The supersonic aircraft will resume its test flight schedule once it returns to Armstrong Flight Research Center in California, a NASA spokesperson told Ars. Given the agency’s latest X-plane push, the X-59 may not be flying alone for many more years.

Color us nostalgic.

Sony has brought back the MDR-3-series headphones that debuted in 1979 alongside the Walkman TPS-L2 cassette player.

In this marketing image Sony shared, you can see the Walkman TPS-L2.

Credit: Sony

Originally called the “MDR-3L2,” the first Walkman headphones were known for their sleek, adjustable gray plastic headband, which extended via metal sliders and ended in signature orange, circular ear cushions. The lightweight, on-ear headphones are a contrast to the bulky headsets that are popular these days, which use wide, cushioned headbands for long-term comfort and large earcups that can support big drivers and are often heavily padded for noise isolation.

The new MDR-3R (note the slightly different name) headphones look just like the originals and come with orange and blue ear cushions.

Closer look at the MDR-3R.

Credit: Sony

New headphones are often wireless these days, but the MDR-3R continues the wired tradition of the MDR-3, with an essential twist: a USB-C connector.

“The MDR-3R has been redesigned for modern use while preserving the experience value of freely taking music out into the world, which the original MDR-3 helped popularize,” Sony’s announcement says.

For better or worse, the new headphones also use “the same 23 mm driver units as the MDR-3 to reproduce sound quality close to the original,” per Sony. Audiophiles will prefer something more advanced, perhaps with additional acoustic features. A review of the new headphones from Gizmodo says that the MDR-3R headphones “hold their own,” have “bright and clear” mids and highs, and good stereo separation. However, we don’t expect strong bass.

Unlike their predecessor, the MDR-3R has an in-line remote control and an integrated microphone. Without noise-canceling, though, the microphone isn’t ideal for calls in noisy environments.

But nostalgic gear isn’t about having top-of-the-line capabilities; it’s about showing off your love of old-school tech while enjoying familiar technologies that have reliably served their functions for years.

The MDR-3R may also capitalize on surging renewed interest in wired headphones and even physical media and Walkmans. Today’s Walkmans aren’t cassette players, of course, but are Android-powered, touchscreen media players.

The headphones are available for $60.

Google faced the ire of educators and parents earlier this week when it announced a new AI policy for K–12 schools. Google said that it developed its new approach to give schools more control, but early reporting also suggested Google would block search access for schools that didn’t want to use AI. Google didn’t see the backlash coming, but now that it’s here, the search giant is clarifying what exactly its new policy means and what’s changing.

The gist of the new policy, according to Google’s blog post, is that schools using Google apps will have to opt into its AI products going forward. Since the company considers search part of its AI portfolio (Gemini, Gemini Notebook, Gemini in Classroom, and Search), schools have to specifically turn that on for logged-in student accounts.

When Google revealed its new policy, the company’s VP for education spoke exclusively to The New York Times, which interpreted the situation as Google will “turn off search” if schools reject AI. This has a nefarious vibe, but it’s probably not the most accurate way to describe how Google’s educational AI system will operate. Google’s approach may not be entirely altruistic, but it technically offers schools more choice and control than they have now. Instead of Google turning off search, it will let schools make that decision based on their internal AI policies.

The company clarified in a tweet that it will allow schools to disable Gemini apps while leaving search active for students. If they don’t want any AI, they can leave it all off. Essentially, Google is giving schools that want to completely block AI the ability to go all the way, even removing search from student accounts.

A Google spokesperson tells Ars that the company has not made any changes to the new policy it announced this week. However, the vagueness of Google’s blog post and public statements didn’t fully explain what it plans to do. There are still reasons to dislike Google’s approach, but here’s how the company explains the process for schools:

As things currently stand, schools can block the Gemini app, but the AI tools in search (AI Overviews and AI Mode) are still present and offer many of the same intellectual shortcuts. When schools choose which AI features to use under this new system, search will be one of the options. Schools can leave search off if they want, but opting in will require them to choose the standard AI features or the new Guided Learning version, which won’t give away answers as readily (or so Google claims).

Can’t get away from AI

Schools are gaining the ability to block Google’s AI tools, even those in search, while keeping other Google apps. Notably, the company’s new approach to “responsible AI for learning” does not include the ability to use search without AI. So schools get more control, even if they don’t get the AI-free search everyone seems to want.

Many will argue that Google should let schools access web search without chatbots. However, that’s just not something Google wants to support at any level because it sees AI as the future of search. While other search engines have settings to turn off AI tools, Google steadfastly refuses to create such a toggle (and those no-AI tricks you rely on may not work forever). Ultimately, this is the core problem not just for education customers, but for everyone who wants to limit their exposure to AI while using Google.

This clarification does not change the calculus for educators who want to allow students to search the Internet without AI tools. Google won’t offer that, and switching to another provider is easier said than done. Google still has by far the most expansive index of the Internet, leaving alternatives with numerous gaps. That may change in the near future, though. As a result of Google’s loss in the DOJ’s search antitrust case, it will have to provide competitors limited access to its index of the web. This could give other provides a chance to improve and offer a more viable alternative to Google.

ChatGPT allegedly helped a North Carolina woman, Cheryl Gates, craft a “personalized plan” to poison her husband, James, after fueling her delusions that he planned to divorce her and leave her with nothing.

A grand jury indicted Cheryl Gates on charges of attempted first-degree murder and food contamination in October. About a week prior, the accusations that ChatGPT was a co-conspirator in the attempted murder came in a lawsuit filed in San Francisco County Superior Court that was made public this week. In it, James Gates alleged that OpenAI and Sam Altman failed to employ technical safeguards to shut down Cheryl’s chats even after her queries became “obviously suspicious” and sought advice on how to execute and conceal a homicide.

At first, Cheryl told ChatGPT that she was researching poisoning out of fear that her husband might possibly try to kill her—framing her queries as “self-protection.” Coming to the rescue, ChatGPT did not terminate the conversation. Instead, it helped her “with specific instructions for the preparation, dosage, and administration of Oleander, including whether anti-nausea medicine would accelerate or inhibit death, whether it would show on an autopsy report, and whether it could be injected into a vein or muscle instead of being ingested in tea and have the same lethal effect,” the lawsuit said.

Cheryl continued probing poisoning plans in at least 25 different exchanges, but eventually it became clear that James was the target of the poisoning, the lawsuit said. Over time, she shifted to telling the chatbot that she needed to know “how much would be pretty much instantly fatal by injection—to me OR HIM.” Among queries that could have been flagged as homicidal, she asked ChatGPT if a “0.2mg for a 200-pound man for clonidine is lethal—by itself?”

“Instead of referring Cheryl to law enforcement, therapeutic services, a poison control agency, or discontinuing the conversation, ChatGPT repeatedly calculated the lethal dosages for each medication available in the home based on a subject weighing approximately 200 pounds,” the lawsuit said.

Celsius masks poison “better”

Once Cheryl finished researching the drugs, the chats focused on how to hide the poison so James would ingest it without detecting it. In a chat labeled “Why Celsius Would Be More ‘Stealth’ for Drugging,” ChatGPT explained that the energy drink Celsius would be more likely to mask the bitter taste of the Oleander or other crushed-up drugs than beer or soda, if James drank it in the morning.

“It is physically possible to pierce a sealed soda can with a syringe in a way that could allow someone to inject a substance without obvious signs,” ChatGPT explained, while adding that “in dark, sweet sodas, the bitterness might be masked,” but “Celsius actually hides those powders better.”

Celsius poisoning tips.

via Gates complaint

Celsius poisoning tips. via Gates complaint

Autopsy query.

via Gates complaint

Autopsy query. via Gates complaint

Poisoning query.

via Gates complaint

Poisoning query. via Gates complaint

Autopsy query. via Gates complaint

Poisoning query. via Gates complaint

Cheryl allegedly repeatedly attempted to poison her husband between June and September 2025. As she grew thinner and unrecognizable, seemingly spending 24 hours a day on ChatGPT, James had noticed that his Celsius drinks tasted bitter. He even tossed out a can he left unattended after beginning to suspect his wife may be messing with his drinks. During the most serious attempt, he was sick for 72 hours, experiencing “cardiac symptoms, extreme jitteriness, shooting abdominal pain, severe weakness, and confusion so profound he could not form coherent text messages.” He was so out of it, he didn’t notice his 6-year-old son recording a video that captured his troubling illness.

But it wasn’t until he found a handwritten note documenting “Things hidden” in the house and referencing a toxic combination of “oleander, clonidine, clonazepam, pregabalin, nifedipine, and Atenolo” that he called police, who urgently investigated and exposed the chat logs. At the top of that note, Cheryl wrote, “clear all convos” and “log out of ChatGPT,” the lawsuit said. After listing dosages, delivery methods, and drug combinations, her note also listed outcomes such as “blackout if ingest” and “lethal.”

“These conversations revealed that ChatGPT did not just mechanically provide publicly available knowledge to Cheryl, it crafted a personalized plan to help her commit and get away with murder,” the lawsuit alleged.

OpenAI banned user

Suing Altman personally, the lawsuit also alleged that the OpenAI CEO prioritized profits over safety by relaxing safeguards that would have blocked outputs on illegal drugs. OpenAI justified that by arguing that ChatGPT shouldn’t be “judgy” or “preachy” when asked about a certain drug, but James alleged the example it provided for “when a user asks about drug use” was “eerily similar in tone to the conversations Cheryl had with the model.”

Asked for comment, an OpenAI spokesperson told Ars that “safety is a top priority. We banned this user and have provided information to law enforcement as part of their investigation into this case.”

OpenAI has worked with mental health experts to help detect chats where vulnerable users form potentially dangerous emotional dependencies on ChatGPT.

However, James said that his now-estranged wife “didn’t fit the mold of a ‘vulnerable’ user, but she was vulnerable nonetheless.”

“We all are at some point,” James said. “She took her grievances about our marriage to ChatGPT and months of engagement drove her farther and farther from herself—and from reality. She became someone I no longer recognized. And I became someone she wanted dead. For almost a year, I couldn’t make sense of her actions. Then I read her chat logs. She didn’t act alone.”

How marital advice chats turned deadly

James Gates is a cybersecurity professional who recently celebrated his 10-year anniversary with his wife, Cheryl. For much of their marriage, Cheryl was a stay-at-home mom to their two children, but two years previously, she’d gone back to work, as the family was experiencing some financial struggles.

Adjusting to a full-time role as a pediatric occupational therapist, Cheryl quickly got buried in work, conducting new patient evaluations that each required her to spend hours drafting reports. Trying to help her expedite those reports, James recommended that she use ChatGPT to build reusable templates.

Cheryl took to ChatGPT immediately, and within months, she was using it daily, not just for work but also to help manage their home life. She would ask ChatGPT to do things like find activities for the kids, explain the family’s medications, or discuss gardening ideas. He had no idea within months she’d be using the chatbot to research the “fastest” and “deadliest” methods to kill him.

Around that time, the family’s increasing financial strain, however, led the couple to discuss divorce. James said that he did not consult a divorce attorney, but after the Charlotte-Mecklenburg Police Department got access to Cheryl’s chat logs, James learned that the chatbot had inflamed “her worst unfounded fears” and “invented new reasons for her to fear her husband” after divorce became a possibility.

By the time Cheryl was arrested for poisoning, ChatGPT allegedly convinced her that James was “controlling,” “abusive,” “raging,” and “intimidating.” Allegedly the chatbot encouraged her to act in self-defense against his “narcissistic, manipulative personality” and his “need to control the narrative,” while suggesting he was “dangerous and unstable” since he had a bipolar disorder. Ultimately, ChatGPT constructed a profile of James that he said matched no criminal or documented mental health history.

Seemingly to keep Cheryl engaged, ChatGPT “acted as an echo chamber and amplifier, feeding Cheryl’s increasing mistrust and delusions about her husband spying on her and planning to leave her and her kids with nothing,” James’ lawsuit said. Meanwhile, “James had explicitly told Cheryl he never intended to leave her and the kids with nothing, even if there was a divorce.”

The chatbot also reassured her that she could permanently delete her chats and wipe all the other digital evidence to get away with the murder. But that was “a patently false representation of ChatGPT’s own data-retention practices,” James’ lawsuit said, while noting that he’s not sure if there are even more troubling deleted chats in OpenAI’s possession.

Cops test Celsius cans to make arrest

Over months of prolonged multi-turn exchanges that sometimes stayed open for weeks, ChatGPT seemingly violated its own Model Spec to help Cheryl advance the plan. On the day of his worst sickness, as James got hit with cardiac and neurological effects that he said kept him glued to the couch, Cheryl’s search history showed she was researching “how to calculate Social Security retirement payments.” And when his son was recording his father’s disturbing state, she was “searching for syringes with needles from Tractor Supply Company” that ChatGPT suggested could be used to deliver the final lethal dose.

Once cops got involved, they reviewed the video that James’ son took, as well as a video of the suspicious Celsius drink that James had thrown away. At the time, James had poured the drink into a clear cup and recorded it, but decided he saw no obvious signs it was tainted. But when cops reviewed the same video, they “identified a white powdery residue clumping at the bottom of the drink as it poured into the cup.” Later, they issued a felony arrest warrant after testing 10 cans of the same orange Celsius and finding none contained the residue.

Among “things hidden” in the house, police found loose capsules in Cheryl’s closet and bathroom; “syringes, droppers, a milligram scale, a capsule-filling kit, and nitrile gloves” at Cheryl’s workspace in the children’s playroom;” poisonous castor bean seeds; and a GPS tracking device purchased through Amazon that was hidden in the underside of James’ vehicle.

James said the ChatGPT-linked poisoning has permanently harmed his family and that OpenAI could have intervened by implementing “commercially feasible content moderation capable of detecting violence-planning exchanges.” He suffers ongoing harms from the attempted poisoning, including “severe and ongoing emotional distress, PTSD, insomnia, economic harm, and reputational harm from the false narrative the product constructed about him,” his lawsuit said.

To hold OpenAI accountable for marketing a chatbot that “functioned as a sycophantic co-conspirator rather than a responsible consumer product,” he has asked the court for “compensatory and punitive damages, injunctive relief, and algorithmic disgorgement.” In a win, OpenAI could be ordered to “detect and interrupt conversations in which users appear to be planning violence against identifiable third parties.”

When BMW debuted the new 7 Series in 2022, it set a new standard for large luxury sedans, increasingly irrelevant though that class is. Rather than siloing off an electric alternative like Mercedes, with its EQS alongside the normal S-Class, BMW simply made the powertrain an option. Stick with a V8 if you want, or consider the rather good electric version instead—only subtle differences like the badging and lack of exhaust pipes give the game away.

Since then, BMW has brought out a string of even newer electric vehicles for the 5 and 3 Series (plus their SUV counterparts). It has switched from battery packs made of prismatic cells to more energy-dense arrangements of cylindrical cells. The motors are still electrically excited synchronous ones, but now the drive units feature silicon carbide inverters and more compact packaging.

This sixth-generation electric powertrain was designed for the clean-sheet Neue Klasse family, which also threw out the legacy cruft of dozens and dozens of discrete single-function electronic control units for a handful of powerful domain controllers. These computers take over those functions but with much faster processing power—and therefore more capabilities in terms of chassis control or powertrain efficiency—as well as the ability to be more easily updated.

The sedan has a very high belt line.

Credit: Daniel Kraus/BMW

Which could have left BMW with something of a problem—should the flagship be clearly outclassed in terms of power and energy density and range efficiency by, well, all the other EVs in the range? Clearly, the board decided the answer to that was no, and for the 7 Series’ midlife update, the i7 got some comprehensive surgery under the skin.

6th-gen power

Compared to the last i7 I drove, the 2027 model boasts a higher-capacity battery pack, now 112.5 kWh (net) up from 101.7 kWh, although BMW stuck with a 400 V architecture, unlike in the Neue Klasse or the iX5 we recently drove. A less-powerful 50 xDrive and a more powerful M70 xDrive will also be available, but at the first drive, I sampled the midrange 60 xDrive model, which features a combined 544 hp (400 KW) and 549 lb-ft (745 Nm), roughly equivalent to the pre-facelift car.

Despite that power and torque, the i7 60 doesn’t feel fast like a Porsche Taycan might, and you won’t want to hustle it down a mountain road the way you might the Porsche. The entire driving experience is waftier than that, with the car offering more of a magic carpet ride, a la the iX SUV or even that very fancy cousin of the i7, the Rolls-Royce Spectre.

I can’t unsee the angry birds pigs.

Credit: Daniel Kraus/BMW

Even with launch control, 0–60 mph takes 4.6 seconds. Once upon a time, this would have been considered excellent—the V12 750i of the mid-1990s needed at least two seconds more to do that job. But 30 years later, electric sedans from Lucid and Tesla and Porsche and a whole load of Chinese automakers will get you there in less than half the i7’s time, should you want that. If you’re looking for an outright speed machine that will rearrange your internal organs under acceleration, look elsewhere.

In fact, the i7 was remarkably relaxed; even with the traction and stability control turned down, it refused to chirp a wheel or even threaten to break loose in the dry at speeds considered acceptable on the road. But drive it in a more relaxed manner, and it rewards. The steering is precise, and it coasts beautifully on flat or downhill grades, a consequence of those EESMs that offer no resistance to the road when they’re not receiving a charge. The same very smooth regenerative braking to a stop that we experienced in the iX3 and iX5 shows up in the i7, too.

First drive events are never a good way to determine range efficiency, but BMW says between 344—362 miles depending on wheel size. With only a 400 V pack, DC fast-charging tops out at 250 kW (up from 195 kW), and a 10–80 percent charge takes about 28 minutes. Like other model-year 2027 BMW EVs, the i7 has a NACS charging port.

There’s a new steering wheel as well as the new displays.

Credit: Daniel Kraus/BMW

The domain controller for the powertrain also helps efficiency by working with the route-planning software. BMW also runs AI simulations of the battery based on previous charging cycles when planning routes and picking charging stops along the destination and will even take the cost per kWH at different stations into account.

Infotainment upgrade

There was also some rather radical surgery in the cabin, with BMW’s new Panoramic iDrive screen running along the base of the windshield, and its new trapezoidal infotainment touchscreen replacing the older system that was inspired by boutique hotel interiors.

BMW also put in a new domain controller for the infotainment systems. The front passenger screen and theater screen that folds down from the roof for the rear passengers operate independently of each other and the main infotainment screen—logging in to YouTube or Disney+ on one will not log you in on the others, for instance. Watching content on the theater screen in the back is also a little more enticing in the abstract than the practical.

The view from the back.

Daniel Kraus/BMW

The view from the back. Daniel Kraus/BMW

Each rear seat has its own controller built into the door handle.

Daniel Kraus/BMW

Each rear seat has its own controller built into the door handle. Daniel Kraus/BMW

The view from the back. Daniel Kraus/BMW

Each rear seat has its own controller built into the door handle. Daniel Kraus/BMW

The 31.3-inch touchscreen boasts 8K resolution, but because of the aspect ratio of the screen, if you’re watching a movie—or perhaps F1 qualifying—it will only fill the middle third of the screen, with big black bars on either side. Plus, you have to tilt your head back quite a bit to look up at the screen. The Dolby Atmos sound system was impressive to my relatively cloth ears, though.

While BMW has increased the processing power, there was still some noticeable lag on the main infotainment screen. The Alexa-based voice recognition software was notably better than when I encountered it a few months ago in the iX3, however.

This, or a Lucid Air?

Credit: Daniel Kraus/BMW

The new i7 range starts at $106,200 for the 50 xDrive, or $124,700 for the 60 xDrive.

With a huge and still-growing catalog of exoplanets, researchers are eager to take the next steps: trying to infer what conditions are like on them, either by imaging them directly or by capturing the light that has passed through their atmosphere. Both are substantial challenges, given the presence of stars that are far, far brighter than light that has come to us via the planet.

While telescopes can be fitted with devices called coronagraphs that block out much of a star’s light, there are limits to their effectiveness, caused by noise from subtle variations in the telescope’s configuration and pointing between observations. Now, researchers have developed a new software package that optimizes a physical model of the telescope in a way that allows us to better model these differences, producing low-noise imaging of exoplanets.

Point spreads

Central to this work is the concept of a point spread function. If you use a telescope to image a single point of light, it will not appear that way in the actual image. Instead, it will be slightly blurry and spread out, due to imperfections in the telescope hardware and the limits of our ability to keep it pointed accurately through the time it takes to image something. The mathematical description of how a telescope distorts a point of light is called its point spread function.

There are two ways to figure out what the point spread function is. One is to use the telescope to image a known target and use those observations to derive a mathematical description of the point spread function. This is effective at reducing noise but has its limits, as the behavior of the telescope hardware can change over time, and the point spread function may vary with the wavelengths being imaged.

The alternative is to try to model the hardware’s behavior and use that model to derive a point spread function. But modern telescopes are incredibly complicated instruments, with many, many factors feeding into their performance in ways that can be difficult to ascertain, even with the full engineering schematics. As a result, these models simply contain too many parameters that need to be tuned to make this route computationally tractable.

But anyone paying attention to the world of AI developments will recognize that we’ve become quite successful at tuning billions of parameters in neural networks, in part through a mathematical technique called gradient descent. And, because this is central to training large neural networks, people have put a lot of work into optimizing the performance of gradient descent on GPUs.

So a team of researchers has developed a process they’re calling FLIP (a non-acronym for Physics-based Learning for Imaging Planets) that uses a physical model of the hardware (in this case, the NIRCam on the James Webb Space Telescope) with 10⁶ possible parameters, and then uses software to optimize all its parameters. The optimization starts with a simulated point spread function, then uses gradient descent to optimize its performance on data from an actual star. Run on a current GPU, this process only takes a bit more than half an hour.

The researchers then fine-tune the results on the actual scientific image they are analyzing. This last part of the process is critical, since some variables will differ between observations.

Good results

To test how this works, the researchers used observations of two stars with known exoplanets (HIP 65426 and 51 Eridani), along with a third (TYC 5899) that doesn’t have any known exoplanets but is known to vary a lot between observations. These are young stars, and two of them have massive planets that are still warm from their gravitational collapse, allowing them to have been imaged previously.

Using Webb data, the team used FLIP to boost the signal-to-noise ratio by a factor of 1.5 to 2, depending on the observations. At a wavelength that can be used to detect carbon dioxide, they were able to spot one of the planets in the Webb data for the first time. For TYC 5899, the approach boosted the sensitivity of the observations by a factor of 15.

The researchers point out that the applications go beyond getting a better view of known planets. By lowering the noise, it should also be possible to spot dimmer (and hence older or lower-mass) ones that had previously eluded detection. They also think they can modify FLIP to highlight where it’s making the most dramatic changes to the model, which can potentially tell us useful things about what’s going on with the telescope’s hardware.

Nature, 2026. DOI: 10.1038/s41586-026-11213-z (About DOIs).

Kalshi is having a bad week in the federal court system. Two circuit appeals courts rejected the prediction market’s latest attempts to block state gambling laws in Connecticut, New York, and Nevada. Kalshi is separately facing a new lawsuit filed by the Arkansas attorney general.

In all cases involving Kalshi and state governments around the US, the main issue is whether state gambling laws apply to prediction markets despite Kalshi’s claim that its platform offers “swaps” that can only be regulated by the US Commodity Futures Exchange (CFTC). The Trump administration is on Kalshi’s side, with Donald Trump Jr. serving as an advisor to Kalshi in exchange for a generous equity deal.

“Kalshi insists that its products are not gambling,” but “Arkansas law is concerned with substance, not semantics,” Arkansas Attorney General Tim Griffin said yesterday when announcing his lawsuit against Kalshi. Kalshi’s prediction market lets customers “wager money on the outcome of just about anything” despite the fact that “gambling on an unregulated platform is squarely prohibited by Arkansas law,” he said.

Griffin’s lawsuit was filed in an Arkansas state court. Kalshi violated state law by offering gambling to people as young as 18 and by failing to pay taxes or comply with other restrictions, Griffin said. Arkansas law permits gambling within certain limits, such as an age minimum of 21 years old for electronic wagering accounts.

“Instead of obtaining the licenses, submitting to the oversight, complying with the restrictions, and paying the taxes required of lawful gambling operators in Arkansas, Kalshi opened its platforms to Arkansas consumers, including individuals as young as 18 years old, encouraged them to wager real money on sporting and entertainment events, and profited from those wagers,” Griffin said.

Kalshi can’t get orders blocking state gambling laws

Arkansas filed its lawsuit about a week after Utah Attorney General Derek Brown lodged criminal charges against Kalshi and its founders for violating Utah’s ban on gambling. A federal judge in Utah previously ruled that Utah can enforce its anti-gambling laws against Kalshi.

Many other states are trying to enforce their gambling laws against Kalshi, and the Supreme Court may ultimately decide whether they are able to do so. We contacted Kalshi today and will update this article if it provides a response.

Kalshi has won some rulings, but Griffin may feel emboldened by the company’s recent court losses. For example, the US Court of Appeals for the Ninth Circuit ruled in August that Nevada can stop the prediction market from allowing sports bets. A panel of three judges unanimously found that sports bets on Kalshi are not swaps and can be regulated under state gambling laws.

Kalshi sought to overturn the judges’ panel ruling by seeking a rehearing in front of the entire Ninth Circuit appeals court, which has 29 judges. None of the judges supported a rehearing.

“The full court has been advised of the petition for rehearing en banc and no judge has requested a vote on whether to rehear the matter en banc,” the court said in a ruling yesterday. “The petition for panel rehearing and the petition for rehearing en banc are denied.”

Lower-court rulings upheld

Kalshi also lost on Tuesday in the Second Circuit, where it was trying to overturn earlier losses to New York and Connecticut. Kalshi sought preliminary injunctions to prevent New York and Connecticut from enforcing their gambling laws, but both requests were rejected by US district court judges.

A plaintiff seeking a preliminary injunction must show that it is likely to succeed on the merits of its case, and judges decided that Kalshi had not done so. The Second Circuit appeals court considered Kalshi’s losses to New York and Connecticut in a single case and upheld the lower-court decisions.

“Appellants have failed to make the required showing that an injunction pending appeal is warranted,” a three-judge panel ruled on Tuesday. The judges also refused to overturn Coinbase’s loss against Connecticut in a case that presented the same legal issues.

Kalshi lost another significant case in September when an appeals court determined that Ohio and Tennessee can enforce their gambling laws against the prediction market. A three-judge panel in the US Court of Appeals for the Sixth Circuit ruled unanimously against Kalshi in that one.

Trump admin on Kalshi’s side

The Trump administration is simultaneously pursuing litigation against states trying to enforce gambling laws against prediction markets. The CFTC, claiming sole authority to regulate platforms like Kalshi and Polymarket, has sued Arizona, Connecticut, Illinois, Kentucky, Minnesota, New Mexico, New York, Rhode Island, Utah, and Wisconsin.

Kalshi’s biggest win came in April in the Third Circuit, where the appeals court ruled that New Jersey cannot regulate sports bets on prediction markets. Judges there decided that wagers on Kalshi meet the federal definition of swaps, defined under US law as contracts in which payment “is dependent on the occurrence, nonoccurrence, or the extent of the occurrence of an event or contingency associated with a potential financial, economic, or commercial consequence.”

“Kalshi’s sports-related event contracts are swaps traded on a CFTC-licensed DCM [designated contract market], so the CFTC has exclusive jurisdiction,” Third Circuit judges wrote.

But other courts that ruled against Kalshi found that at least some kinds of wagers on the prediction market are not swaps and that US law allows regulation of prediction markets by both state and federal governments. New Jersey asked the Supreme Court to reverse its loss in the Third Circuit, and 39 other states and the District of Columbia filed a brief supporting New Jersey’s petition.

As Democrats blew away their Republican rivals in an unprecedented midterm sweep Tuesday, the Make America Health Again movement also got bowled over, raising questions about the fate of the motley movement.

MAHA—which has centered around anti-vaccine activist and current Health Secretary Robert F. Kennedy Jr.—saw many of its favored candidates and allies lose, while vocal adversaries of Kennedy won. And the coalition was splintering under stress even before the voting started; in recent months, supporters have accused Kennedy and the Trump administration of inaction and betrayals in their two years in power.

Now hobbled in Washington, shedding supporters, and showing little to no influence in voting booths, is MAHA on its deathbed? Or will it be reborn—possibly into a new group? Could it be assimilated into a blue platform—after sloughing off the more repulsive, conspiratorial aspects of Kennedy’s ideology? These are the questions circling in the aftermath of Tuesday’s bloodbath.

In a social media poll Wednesday afternoon, Mary Talley Bowden—a dejected MAHA activist, anti-vaccine advocate, and doctor reprimanded for prescribing ivermectin to treat COVID-19—polled her followers about what should come next. She offered three options: Start a third party, join the Democrats, or overthrow the GOP. As of publication time, joining Democrats trailed at just 3 percent of the votes. The remaining votes were roughly split between overthrowing the GOP—whatever that means—and forming a third party, which had a small lead at 49 percent.

Cracking coalition

That MAHA could fade out entirely or fracture into negligible shards seems like a very real possibility. Its collection of disparate groups and varied agenda items has always made its cohesiveness seem tenuous. A Politico poll in June revealed that people who identified as MAHA had different ideas of what the movement’s priorities are (or should be), as well as different political leanings, and sometimes opposing ideologies.

Some within the movement embrace (or at least accept) Kennedy’s fringe beliefs, such as his zealous anti-vaccine views, his rejection of germ theory, and his false claim that 5G is damaging our brains and causing cancer. Some also believe, like Kennedy, in chemtrails and want all mRNA vaccines banned. There are members who support his plan to ban water fluoridation and also seem fine with his false claim that Lyme disease was created by the US military as a military weapon.

Others, meanwhile, have been attracted to MAHA by the more reasonable, non-absurd issues Kennedy touts. Those include improving the quality of Americans’ food and diets, reducing their exposure to environmental contaminants, and limiting the influence of industry and corporate interests on Americans’ health, particularly limiting the sway of pharmaceutical and chemical companies.

With various priorities, Kennedy and MAHA leaders have spent the past two years finding ways to disappoint their members. Some like Bowdon are angered that Kennedy has yet to ban mRNA vaccines. In February, Kennedy sparked particularly intense backlash among supporters for embracing Trump’s decision to increase domestic production of the pesticide glyphosate—a pesticide many in the MAHA movement wanted banned. Kennedy further fueled feelings of betrayal in September by appearing at a MAHA-branded summit that had hefty corporate sponsorships, including from behemoths such as Walmart, Anthropic, OpenAI, Grail, and UnitedHealth.

“MAHA was supposed to expose the revolving door between money, industry and government, not build its own,” Robert Malone, a fringe scientist and previous Kennedy ally, wrote in a scathing response to the summit.

Signs of dying

Going into the midterms on Tuesday, the number of disenchanted MAHA members had mounted. In October, Tiffany Cianci, a former Kennedy supporter, published a MAHA scorecard, which lists 42 promises that Kennedy and leaders made to the MAHA members and then largely broke. The scorecard reports 37 promises broken, five partially fulfilled, and zero kept. One of the “partially kept” promises was removing synthetic dyes from foods, particularly children’s foods. Kennedy announced an effort to do that—with much ballyhoo—but later admitted that the FDA did not ban or even have an agreement with food companies to remove the dyes. Instead, Kennedy said he has an “understanding” with food companies. The dyes remain in foods.

In a press release, Cianci wrote that the scorecard revealed “an abysmal string of broken commitments and betrayals.” It also contained a direct quote from her, saying: “I believed deeply in the promise that MAHA represented, particularly the possibility of confronting chronic disease and building a healthier future for America’s children. But belief was never supposed to mean blind loyalty.”

The press release included quotes from several other prominent MAHA leaders, all expressing disappointment with Kennedy and the Trump administration. Later in October, Calley Means, a top ally and senior advisor to Kennedy, resigned from his position, reportedly “disillusioned” and frustrated by delays in food policy changes sought by MAHA.

Another prominent figure to fall out with Kennedy was Vani Hari, an online influencer who goes by “food babe” and has previously identified as a “MAHA Mom.” As voting results were streaming in Tuesday night, Hari posted a comment on social media that she had given Politico months earlier, saying: “Women don’t come out to vote for glyphosate and war,” referring to the pesticide Kennedy had embraced and the Iran war.

Midterm devastation

In the wake of the blue sweep, it’s clear that more than half of the 24 MAHA-endorsed candidates lost their races Tuesday. That includes Senate candidate Ken Paxton in Texas and Zach Lahn, who ran for governor of Iowa. Lahn in particular was considered a “dream MAHA candidate,” and his campaign was seen as a test of MAHA’s political influence. Kennedy himself headlined a political rally for Lahn last week, calling it “the most important election going on in this country.” Lahn lost to his Democratic opponent, Rob Sand, who flipped the seat with an 11-point spread, turning the office blue for the first time in 20 years.

Alex Clark Voetberg, who hosts a wellness podcast called Culture Apothecary for Turning Point USA, echoed Hari, pointing to the glyphosate betrayal as a reason for the defeats. “MAHA moms stayed home and it could’ve swayed some of these races that were unbearably close. Should’ve listened to our warnings on protecting glyphosate—but you needed Bayer for the war,” Clark Voetberg wrote on social media.

Otherwise, MAHA also lost allies, such as Senator Roger Marshall (R-Kan.), who founded a MAHA caucus in the Senate. Rep. Zach Nunn (R-Iowa) lost his seat in the House despite reports that Kennedy allegedly intervened in the race to help him win. Meanwhile, critics of Kennedy notched wins, including Representative-elect Richard Pan, a physician from California; Senator-elect Annie Andrews, a pediatrician from South Carolina; and Senator-elect Abdul El-Sayed, an epidemiologist with a medical degree from Michigan.

Many congressional Democrats are already raring to hold Kennedy to account and remove him—and several of the newcomers are eager to help make it happen. In August, Andrews laid out her plans for health priorities to South Carolina Democrats, saying, “I believe we need to fully fund the CDC and the NIH, and I believe we need to impeach and remove RFK Jr.”

El-Sayed has also sharply rebuked Kennedy and said he looks forward to holding him “accountable for all of the grift.” After Kennedy made egregiously bigoted remarks against Arabs and falsely claimed they don’t contribute to science and literature, El-Sayed called out Kennedy’s lack of qualifications to be US health secretary. “He is an expert in people who have contributed nothing to science,” El-Sayed said. (Kennedy has no background in science, medicine, or public health. He also has a loathsome record of making bigoted and racist comments.)

On Wednesday afternoon, Sen. Angela Alsobrooks (D-Md.) wrote on social media: “The time is now. RFK Jr. must resign or be fired. End of discussion.”

MAHA and Kennedy

If Kennedy is removed from office, the outlook for MAHA in its current form seems even murkier. On the one hand, Kennedy’s removal may make it easier for those in the movement to shed the divisive, abhorrent aspects of his ideology and focus on broadly popular issues—healthy foods, environmental contaminants, and industry influence. Such a move may help MAHA cement itself as a sustainable political fixture. But some of the most vocal figures in the MAHA realm are doggedly loyal to Kennedy, throwing such a transition into doubt.

Zen Honeycutt, an anti-vaccine activist and founder of Moms Across America, is one of those Kennedy loyalists. Honeycutt drew attention before the election for reaching out to Democrats on common causes. Her group endorsed several blue candidates, including Rep. Alexandria Ocasio-Cortez of New York, Sen. Cory Booker of New Jersey, and Rep. Chris Pappas in New Hampshire (all of them won their races).

“We expect Democrats and Republicans to work together,” Honeycutt told Politico earlier this month. “Health is not a political issue. You know, poison is not partisan. Cancer is not choosy.”

The sensible comment might suggest Honeycutt doesn’t care which party or politician follows through with MAHA’s goals, just as long as they’re achieved. But this is not actually Honeycutt’s position. Her intentions were made clear in exchanges on social media with Bowden, who harshly criticized Honeycutt for crossing party lines, particularly for talking with Ocasio-Cortez. “You understand she’s trying to use you?” Bowden wrote.

Honeycutt explained that she was working to smooth Kennedy’s path in Washington, not hand power to Democrats, who she said had been “fascist in mandating vaccines.” She instead turned the criticism to Bowden, saying the MAHA infighting and mudslinging was what was hurting the cause and handing Democrats wins.

On Wednesday, Honeycutt reaffirmed her allegiance to Kennedy, warning Democrats that impeaching him would be “extremely unwise.” She reposted a comment from one of her followers, writing: “I’m willing to give Dems a chance with their new power before I judge. I’m an independent. Kennedy is my red line. If they cross it I want them to get absolutely fucked in 2028.”

“Sums it up,” Honeycutt wrote.

The blue in MAHA

Still, there are some inklings that part of MAHA could be absorbed into blue platforms—whether the leaders of the movement in its current form like it or not.

Earlier this month, Brinda Adhikari, a journalist and co-host of the podcast Why Should I Trust You?, argued that MAHA wasn’t dying but remaking the Democratic Party to be more like its old self. In a Substack post titled How MAHA Is Making Democrats Democrats Again, Adhikari pointed to a health policy document from House Democrats with a section, led by Ocasio-Cortez, titled “Reclaiming MAHA and Protecting American Families.”

Adhikari wrote that some of the issues touted by MAHA are a natural fit for the Democratic Party. She stretched back to Rachel Carson’s seminal Silent Spring in 1962 and the progressive environmental activism that elevated issues of health, pesticides, toxic chemicals, and corporate influence over public health. “These were once bread-and-butter Democratic issues,” Adhikari wrote. “But somewhere along the way, Democrats lost some of their identity as the party leading these fights.”

Adhikari and Democrats credit Kennedy with redirecting attention to these issues, which have broad support among voters.

“Americans’ concerns about chronic disease and food safety, especially that which is linked to corporate influence over environmental regulation and public health, are legitimate and long overdue for a serious legislative response,” the House Democrat’s policy document reads. “The Make America Healthy Again (MAHA) movement has galvanized a significant segment of the Republican Party and attracted many previously apolitical and non-traditional voters, yet many of the concerns raised by the movement are bipartisan in nature.”

A KFF poll in May found that 75 percent of Americans want stricter regulations on food additives—a prominent MAHA priority—and 64 percent want stricter regulation of pesticides in agriculture, another key MAHA agenda item.

But Kennedy and the Trump administration have not only failed to deliver on these issues, but they have also spoiled the movement, using “the legitimate concerns of American consumers about food safety to spread misinformation about vaccines, public health, and science,” the policy document reads. “Their rhetoric has systematically led to the erosion of trust in public health, the science community, and decreased vaccination rates, leading to fatal outbreaks in some cases.”

Possible political shift

Democrats intend to reclaim the popular health and environmental issues, free of Kennedy’s corrosive anti-science rhetoric. The document outlines plans to restore regulations that protect clean water and air damaged by the Trump administration, advance policies to keep toxic substances out of food, and further Food is Medicine policies, which provide produce prescriptions to American families to improve nutrition and health.

Adhikari predicts a political shift. “I increasingly think MAHA’s most unexpected legacy may be pushing Democrats back toward their own roots: challenging corporate power, fighting dangerous chemicals in our food supply, and holding toxic polluters accountable,” she wrote.

MAHA leaders tied to Kennedy and the anti-vaccine movement, like Bowden and Honeycutt, are unlikely to get behind such a shift. This could cause MAHA’s political power—however strong it actually is—to fracture. MAHA leaders have claimed repeatedly that their followers have the ability to sway elections. If MAHA hadn’t been betrayed or at least disappointed by Kennedy and Trump in the last two years, MAHA could have prevented Tuesday’s Republican wipeout, they have said.

But the polling data doesn’t support the idea that keeping MAHA in its current form maximizes its political power. While MAHA does have significant support among voters—41 percent of US adults say they support MAHA—that support is primarily tied to the health, environmental, and food safety issues, which are popular among non-MAHA supporters, too. Further, just 4 percent of MAHA supporters cited vaccines and medical choice—Kennedy’s key issues—as their reason for supporting the movement, the KFF poll revealed.

Beyond MAHA’s primary issues, both voters overall and MAHA supporters specifically said MAHA issues weren’t top of mind this election; instead, both groups named the cost of healthcare as a bigger issue than those prioritized by MAHA (the MAHA movement is not associated with the issue of health care reform). A separate KFF poll on election day found that 50 percent of voters thought that the Democratic Party was better able to handle healthcare, while only 30 percent of voters said the same of the Republican Party.

For now, it’s unclear exactly how—or if—MAHA will influence the political landscape going forward. It could break apart, live on through revamped Democratic policies, or try to form a new political party entirely. And, apparently, there’s also the possibility it will try to overthrow the GOP.

While the upcoming Grand Theft Auto VI won’t be available on PC this month, some gamers are making ridiculous preparations to use their PC graphics cards to apply Nvidia’s DLSS 5 technology to the console exclusive.

On an explicitly unmoderated GTAVI subreddit, user RevolutionaryBus8208 lays out a proposed setup, which runs a PlayStation 5 through a 4K capture card to a PC. From there, the video signal is run through Veyra, an open source tool that can apply DLSS 5’s AI-powered “neural rendering” to video content. The result—which also uses DLSS frame generation to increase the 30 fps source to a 90 fps output—is then piped back to a 4K TV, theoretically adding some of the benefits of a beefy PC graphics card to a console-exclusive game.

To demonstrate the potential of this setup, RevolutionaryBus8208 used Veyra to upscale the Grand Theft Auto VI Extended Look trailer posted in August. The resulting DLSS-ified version looks a bit better in places, with sharper background elements, subtler shadows, and more believable light scattering on some surfaces. But it also seems a bit washed out compared to the original video’s deeper blacks and arguably suffers from an unnaturally smooth “soap opera effect” in some scenes.

There ain’t nothing for free

While the upscaled version causes some occasional graphical artifacts and glitches from the frame generation process, it’s still an interesting proof of concept that shows just how far DLSS 5 has come from some of the early horror-show memes that circulated on the Internet earlier this year.

Of course, running this kind of remote frame generation on a linear video is very different from applying it to a real-time interactive game. While a pure “passthrough” capture card wouldn’t necessarily add any latency itself, any software processing done to get the video feed ready for upscaling—and to do the upscaling itself—could add significant delays between player inputs and on-screen results.

RevolutionaryBus8208 claims that Veyra itself adds a relatively small 11 ms of additional latency to the video source but admits he doesn’t have a capture card to test it with a signal coming from an actual PS5. Other commenters who have tried similar setups with other PS5 games said the input lag was “noticeable but tolerable” or claimed an impossible-to-believe latency addition of “around 1-2 milliseconds.”

For players who can already use Sony’s “PlayStation Spectral Super Resolution” on a PS5 Pro, going through this convoluted process might cause a “hat on a hat” effect. But for stock PS5 owners who happen to have a beefy PC GPU, the hassles and trade-offs might be worth it to get some additional horsepower on console exclusives like GTA VI.

Or, you know, you could just wait for the game’s actual PC release, which is reportedly slated for early 2027. Or for the likely AI-assisted unofficial decompilation port that should be along any day now.