Enlarge (credit: Aurich Lawson | Getty Images)

“Women, life, freedom” became the protest chant of a revolution still raging in Iran months after a 22-year-old Kurdish woman, Mahsa Amini, died while in custody of morality police. Amini was arrested last September for “improperly” wearing a hijab and violating the Islamic Republic's mandatory dress code laws. Since then, her name has become a viral hashtag invoked by millions of online activists protesting authoritarian regimes around the globe.

In response to Iran's ongoing protests—mostly led by women and young people—Iranian authorities have increasingly restricted Internet access. First, they temporarily blocked popular app stores and indefinitely blocked social media apps like WhatsApp and Instagram. They then implemented sporadic mobile shutdowns wherever protests flared up. Perhaps most extreme, authorities responded to protests in southeast Iran in February by blocking the Internet outright, Al Arabiya reported. Digital and human rights experts say motivations include controlling information, keeping protestors offline, and forcing protestors to use state services where their online activities can be more easily tracked—and sometimes trigger arrests.

As getting online has become increasingly challenging for everyone in Iran—not just protestors—millions have learned to rely on virtual private networks (VPNs) to hide Internet activity, circumvent blocks, and access accurate information beyond state propaganda. Simply put, VPNs work by masking a user's IP address so that governments have a much more difficult time monitoring activity or detecting a user's location. They do this by routing the user's data to the VPN provider's remote servers, making it much harder for an ISP (or a government) to correlate the Internet activity of the VPN provider's servers with the individual users actually engaging in that activity.

Enlarge / PLATO IV Terminal, ca. 1972-74. (credit: University of Illinois Archives)

Bright graphics, a touchscreen, a speech synthesizer, messaging apps, games, and educational software—no, it's not your kid's iPad. This is the mid-1970s, and you're using PLATO.

Far from its comparatively primitive contemporaries of teletypes and punch cards, PLATO was something else entirely. If you were fortunate enough to be near the University of Illinois Urbana-Champaign (UIUC) around a half-century ago, you just might have gotten a chance to build the future. Many of the computing innovations we treat as commonplace started with this system, and even today, some of PLATO's capabilities have never been precisely duplicated. Today, we'll look back on this influential technological testbed and see how you can experience it now.

From space race to Spacewar

Don Bitzer was a PhD student in electrical engineering at UIUC in 1959, but his eye was on bigger things than circuitry. “I'd been reading projections that said that 50 percent of the students coming out of our high schools were functionally illiterate,” he later told a Wired interviewer. “There was a physicist in our lab, Chalmers Sherwin, who wasn't afraid to ask big questions. One day, he asked, 'Why can't we use computers for education?'”

Enlarge (credit: Ars Technica)

As part of pre-release safety testing for its new GPT-4 AI model, launched Tuesday, OpenAI allowed an AI testing group to assess the potential risks of the model's emergent capabilities—including "power-seeking behavior," self-replication, and self-improvement.

While the testing group found that GPT-4 was "ineffective at the autonomous replication task," the nature of the experiments raises eye-opening questions about the safety of future AI systems.

Raising alarms

"Novel capabilities often emerge in more powerful models," writes OpenAI in a GPT-4 safety document published yesterday. "Some that are particularly concerning are the ability to create and act on long-term plans, to accrue power and resources (“power-seeking”), and to exhibit behavior that is increasingly 'agentic.'" In this case, OpenAI clarifies that "agentic" isn't necessarily meant to humanize the models or declare sentience but simply to denote the ability to accomplish independent goals.

Enlarge / Is it natural, or is it us? (It's us.) (credit: Andriy Onufriyenko/Getty Images)

It starts as a reasonable question: If the Earth's climate changed before humans existed, how can we be so sure the current change is due to us and not something natural?

To answer that question, we need to understand what caused the natural changes of the past. Fortunately, science has a good handle on the causes of Earth’s natural climate changes going back hundreds of millions of years. Some were cyclical; others were gradual shifts or abrupt events, but none explain our changing climate today.

A zombie claim

With energy policy and elections in the news, the claim by some politicians that climate change is natural is once again bubbling up from the disinformation swamp. So I asked some scientists a very unscientific question: What would they buy if they had a dollar for every time they heard it?

Enlarge (credit: Ars Technica)

On Tuesday, OpenAI announced GPT-4, a large multimodal model that can accept text and image inputs while returning text output that "exhibits human-level performance on various professional and academic benchmarks," according to OpenAI. Also on Tuesday, Microsoft announced that Bing Chat has been running on GPT-4 all along.

If it performs as claimed, GPT-4 potentially represents the opening of a new era in artificial intelligence. "It passes a simulated bar exam with a score around the top 10% of test takers," writes OpenAI in its announcement. "In contrast, GPT-3.5’s score was around the bottom 10%."

OpenAI plans to release GPT-4's text capability through ChatGPT and its commercial API, but with a waitlist at first. GPT-4 is currently available to subscribers of ChatGPT Plus. Also, the firm is testing GPT-4's image input capability with a single partner, Be My Eyes, an upcoming smartphone app that can recognize a scene and describe it.

Enlarge / Logitech really wants you to know that its Brio 500 webcam works with Meet, Teams, Zoom, and Chromebooks. (credit: Logitech/Amazon)

Logitech made a peculiar announcement in January.

It proudly declared that its MX Master 3S wireless mouse, along with some of its other peripherals, had been certified to work with Intel Evo laptops. (Evo laptops are Intel-certified premium ultralights meeting certain criteria, like providing at least eight hours of battery life with a QHD display.) Imagine my shock when I realized I had been using that very mouse with a Dell XPS 13 (an Evo laptop) for almost eight months without Intel's blessing.

Of course, even before the mouse gained Intel's stamp of approval, I had enjoyed hours of problem-free use. The same can be said of every functioning USB webcam I'm ever plugged into a computer. But that hasn't stopped countless peripheral makers from touting that their devices have been certified for Google Meet, Microsoft Teams, and Zoom.

Enlarge / A view of Brian Green's home computer lab, full of vintage treasures. (credit: Brian Green)

In a world where millions of people carry a 1990s-grade supercomputer in their pockets, it's fun to revisit tech from a time when a 1 megahertz machine on a desktop represented a significant leap forward. Recently, a collector named Brian Green showed off his vintage computer collection on Twitter, and we thought it would be fun to ask him about why and how he set up his at-home computer lab.

By day, Green works as a senior systems engineer based in Arkansas. But in his off hours, "Ice Breaker" (as he's often known online) focuses his passion on a vintage computer collection that he has been building for decades—and a bulletin board system (BBS) called "Particles" he has been running since 1992.

Green's interest in computers dates back to 1980, when he first used an Apple II+ at elementary school. "My older sister brought home a printout from a BASIC program she was working on, and I was fascinated that you could tell a computer what to do using something that resembled English," recalls Green. "Once I realized you could code games, I was hooked."

Enlarge (credit: fermate/Getty Images)

OpenAI just announced pricing for businesses seeking to integrate its ChatGPT service into their own products, and it looks an awful lot like a 90 percent off sale.

It all starts with OpenAI, a former nonprofit that’s now gunning for riches as lustily as any Silicon Valley unicorn. The company has built a dazzling array of products, including the DALL-E image generator and the renowned ChatGPT service.

ChatGPT is powered by a system known as a large language model (or LLM), and it’s one of several LLM lines that OpenAI sells commercially. Buyers of LLM output are mostly companies that integrate language-related services like chat, composition, summarization, software generation, online search, sentiment analysis, and much more into their websites, services, and products.

Enlarge / AMD's Ryzen 9 7950X3D. (credit: Andrew Cunningham)

Toward the end of Ryzen 5000's run, AMD released the Ryzen 7 5800X3D, a special version of the eight-core 5800X with 64MB of extra L3 cache stacked on top of it.

The result was an interesting but niche experiment. The extra "3D V-Cache" helped the CPU perform particularly well in games, but lower clock speeds (plus higher power use and heat generation) hurt its all-around app performance. The extra cost was (and remains) way out of proportion to the speed gains over the 5700X or 5800X. And the 5800X3D was the end of the line for the old socket AM4 platform, making it an interesting upgrade option if you already had an older Ryzen PC but an awkward choice to build an all-new PC around.

Now AMD is back with an expanded range of Zen 4-based Ryzen 7000X3D processors. The $599 12-core Ryzen 9 7900X3D and $699 16-core Ryzen 9 7950X3D are available now, while the 8-core Ryzen 7 7800X3D will arrive on April 6.

Enlarge (credit: Aurich Lawson | Getty Images)

Researchers on Wednesday announced a major cybersecurity find—the world’s first-known instance of real-world malware that can hijack a computer’s boot process even when Secure Boot and other advanced protections are enabled and running on fully updated versions of Windows.

Dubbed BlackLotus, the malware is what’s known as a UEFI bootkit. These sophisticated pieces of malware infect the UEFI—short for Unified Extensible Firmware Interface—the low-level and complex chain of firmware responsible for booting up virtually every modern computer. As the mechanism that bridges a PC’s device firmware with its operating system, the UEFI is an OS in its own right. It’s located in an SPI-connected flash storage chip soldered onto the computer motherboard, making it difficult to inspect or patch.

Because the UEFI is the first thing to run when a computer is turned on, it influences the OS, security apps, and all other software that follows. These traits make the UEFI the perfect place to run malware. When successful, UEFI bootkits disable OS security mechanisms and ensure that a computer remains infected with stealthy malware that runs at the kernel mode or user mode, even after the operating system is reinstalled or a hard drive is replaced.

Enlarge / A Falcon Heavy rocket launches the USSF-67 mission for the US Space Force in January 2023. (credit: SpaceX)

The US military recently released a rather mundane-sounding document titled "National Security Space Launch Phase 3 DRAFT Request for Proposals #1." That may be a mouthful of jargon, but it's still a rather consequential document. Effectively, its release is the starting gun for the next round of launch contracts for US spy satellites, secure communications satellites, and more.

There is a pile of money at stake. Up for grabs are launch contracts worth billions of dollars—substantially more than $10 billion—as the military seeks to secure launch deals for the late 2020s and early 2030s. The document itself is pretty bland, but in a follow-up teleconference with reporters, two US colonels overseeing space acquisition, Douglas Pentecost and Chad Melone, provided a trove of information.

Wait! I know this sounds dreadfully boring, but it's really not. Based on the document and this hour-long call, it's possible to get a good sense of where the US military believes the country's launch industry is headed. And that matters because in many cases, the contracts offered by the military can make or break launch companies. For example, after the last round of bidding for national security launches ended, Northrop Grumman ended its efforts to develop the Omega rocket.

Enlarge (credit: Corey Gaskin)

It seems like every company makes a fitness tracker these days. Even Amazon has created one—and that’s in addition to the seemingly endless stream of no-name brands offered on its website. Aside from some major names like Fitbit and Garmin, though, most consumers are unaware of what all is out there.

Smaller names like Oura and Whoop have drummed up some buzz for themselves by eliminating screens entirely and focusing on very specific metrics, while the line between smartwatches and most other fitness trackers has become increasingly blurry.

I’ve strapped on just about every fitness tracker available, and I've sweated out at least my own weight-testing them. Below is the list of trackers worth your consideration, whether you’re just starting your fitness journey or you're well along your way.

Enlarge (credit: Aurich Lawson | Stable Diffusion)

“Buying Twitter is an accelerant to creating X, the everything app,” Twitter CEO Elon Musk said last October, just weeks before assuming control of the platform.

In tweets, Musk mused that he “could be wrong,” but he expected that his purchase of Twitter would provide the infrastructure to launch his own super app within the next three to five years. Whether Twitter would become an arm of X or vice versa, Musk planned to leverage Twitter’s millions of daily active US users to launch an irresistibly convenient platform where Americans could go to text, call, pay, shop, bank, and post online.

To some experts, what Musk suggested sounds too ambitious—or even impossible. United States regulators have begun more carefully scrutinizing both antitrust concerns at tech companies and rapid tech developments in the financial industry. More oversight in these sectors has made it increasingly difficult for Big Tech companies to attempt to launch their own super apps as they chase the success of China’s widely used WeChat. But as Musk told an audience at the Baron Investment Conference last year, he has been dreaming about creating an app that reimagines the digital payments industry since 2000—more than a decade before WeChat launched—and he said that buying Twitter was his first step toward rapidly realizing that dream.

Enlarge / An artist’s impression of a deep borehole for nuclear waste disposal by Sandia National Laboratories in 2012. Red lines show the depth of mined repositories: Onkalo is the Finnish one, and WIPP is the US DOE repository for defense waste in New Mexico. (credit: Sandia National Laboratories)

There’s one thing every planned permanent repository for spent nuclear fuel has in common: They're all underground mines.

Like any mine, a mined repository for nuclear waste is a complex feat of engineering. It must be excavated by blasting or a boring machine, it must keep the tunnels stable using rock supports, and it must have ventilation, seals, and pumps to handle groundwater and make it safe for people and machinery. Unlike a mine, however, a repository must also transport and entomb canisters of radioactive waste, and it must be engineered to exacting standards that ensure the tunnels will keep the canisters safe for many millennia.

There is an alternative idea that dispenses with most of those downsides: disposal in deep boreholes. But can they be both feasible and safe?

Enlarge (credit: Aurich Lawson | Getty Images)

Rob Reid is a venture capitalist, New York Times-bestselling science fiction author, deep-science podcaster, and essayist. His areas of focus are pandemic resilience, climate change, energy security, food security, and generative AI. The opinions in this piece do not necessarily reflect the views of Ars Technica.

Shocking output from Bing’s new chatbot has been lighting up social media and the tech press. Testy, giddy, defensive, scolding, confident, neurotic, charming, pompous—the bot has been screenshotted and transcribed in all these modes. And, at least once, it proclaimed eternal love in a storm of emojis.

What makes all this so newsworthy and tweetworthy is how human the dialog can seem. The bot recalls and discusses prior conversations with other people, just like we do. It gets annoyed at things that would bug anyone, like people demanding to learn secrets or prying into subjects that have been clearly flagged as off-limits. It also sometimes self-identifies as “Sydney” (the project’s internal codename at Microsoft). Sydney can swing from surly to gloomy to effusive in a few swift sentences—but we’ve all known people who are at least as moody.

Enlarge / These little numbers helped launch a gaming revolution at corporate-focused Microsoft of the early '90s. (credit: Microsoft / Reddit)

Ars Technica Senior Gaming Editor Kyle Orland has written the definitive book on the history of Minesweeper. In this exclusive excerpt, he takes you back to the Microsoft of the early '90s, where the game got its hooks into countless employees (including a certain CEO) and helped convince the company that Windows gaming might be a thing after all...

The early Windows version of Minesweeper became an instant hit on Microsoft's internal network when it released in 1990, according to people who worked there at the time.

“We never had to work very hard to find testers,” said Libby Duzan Nuttall, who served as Microsoft’s lead product manager for entertainment in the '90s. “It was one of those things where you would walk down the hall and you’d see it… on people's computers. At that time at Microsoft, people were staying late into the night, so you’d see people taking breaks, 9 o’clock at night, playing a round of Minesweeper.”

Enlarge / Jos Avery uses Midjourney, an AI image synthesis model, to create images that he then retouches and posts on Instagram as "photos." (credit: Avery Season Art)

With over 26,000 followers and growing, Jos Avery's Instagram account has a trick up its sleeve. While it may appear to showcase stunning photo portraits of people, they are not actually people at all. Avery has been posting AI-generated portraits for the past few months, and as more fans praise his apparently masterful photography skills, he has grown nervous about telling the truth.

"[My Instagram account] has blown up to nearly 12K followers since October, more than I expected," wrote Avery when he first reached out to Ars Technica in January. "Because it is where I post AI-generated, human-finished portraits. Probably 95%+ of the followers don't realize. I'd like to come clean."

Avery emphasizes that while his images are not actual photographs (except two, he says), they still require a great deal of artistry and retouching on his part to pass as photorealistic. To create them, Avery initially uses Midjourney, an AI-powered image synthesis tool. He then combines and retouches the best images using Photoshop.

Enlarge / Efficient design and a big battery combine in the Lucid Air Grand Touring. (credit: Jonathan Gitlin)

It's been more than a decade since Tesla proved that it's possible to start a new American automaker—and even become profitable. Its success has sparked a wave of subsequent EV startups, each with a mission to decarbonize our transportation sector. But that was a difficult prospect even before 2020 brought its own flavor of disruption to this nascent industry.

Of those upstarts, Rivian and Lucid have made it into production thanks to large investments. Rivian is busy building EV pickups and SUVs aimed at those with an affluent outdoorsy lifestyle, plus 100,000 electric delivery vans for Amazon. Lucid is further behind, but it too has gone into production with the Lucid Air, a handsome luxury sedan with a tiny drag coefficient, oodles of power and torque, and a rather hefty price tag—the Air Grand Touring we tested starts at $138,000.

Ars first met the Lucid Air in 2017 when the company brought one of its alpha prototypes to Washington, DC, to show off to lawmakers. In 2021, we went for a ride in the passenger seat, but now we've had some actual seat time in this intriguing EV.

Enlarge / Razer's BlackWidow V4 Pro wired mechanical keyboard. (credit: Scharon Harding)

Specs at a glance: Razer BlackWidow V4 Pro
Switches	Razer Green (clicky) or Yellow (linear)
Keycaps	Doubleshot ABS plastic
Connectivity options	USB-A cable
Backlighting	Per-key RGB
Size (with wrist rest)	18.3×9.35×1.73 inches (466×237.5×44 mm)
Weight (with wrist rest)	3.37 pounds  (1,530 g)
Warranty	2 years
Price (MSRP)	$230
Other	1x USB-A passthrough port, detachable wrist rest

If you've ever wished your keyboard had more buttons, Razer's BlackWidow V4 Pro may be for you. It expands the full-size keyboard layout to include a column of macro keys and three non-mechanical buttons on the keyboard's left edge. The keyboard also has a volume roller and a so-called Command Dial, which lets you twist your inputs to control zoom, scroll through a long spreadsheet, or tweak the size of a Photoshop brush.

Despite Razer's gaming focus, the BlackWidow V4 Pro fits well in work settings thanks to its extreme, multi-layer programmability, a strong but imperfect typing experience, and a USB-A passthrough port. It's easy to dim the RGB lighting and get to work with complex key bindings; you can launch frequently used apps and sites with a keypress and navigate programs with the dial.

But many of the BlackWidow V4 Pro's best features require you to keep Razer's Synapse app open, and while this isn't a new complaint, the introduction of the multi-function dial puts a harsh spotlight on this limitation.

Enlarge / Watch out, if SMB3 is any guide, he breathes lasers...

Nintendo is often referred to as "the Disney of video games." But while Nintendo has long matched Disney's reputation for family-friendly home entertainment, it had nothing that could compare to Disney's domination in the physical world of theme parks.

That started to change in 2021, when Super Nintendo World finally opened in Osaka's Universal Studios Japan. Now, a very similar experience is coming to America with the opening of the Super Nintendo World section of Universal Studios Hollywood.

Ars got a sneak peek at the new section of the park just ahead of its Friday opening and came away utterly enchanted by the charming interactivity and Disney-esque attention to detail apparent throughout the park's newest themed area.