The terms "blindingly obvious," "logical consequence," and "that is not how it works" appear nowhere in the government handbook of internet legislation. In particular, the discovery that imposing age access controls on websites has pushed users to VPNs has come as a huge surprise to legislators in the UK, the EU, Canada, and Australia. Nobody here knows how old VPN users are, be they kids unwilling to lose access or adults unwilling to disgorge personally identifying data to who knows what.

As they recover from this shocking discovery, these fine people are looking at ways to control VPNs, whether by adding age verification here too or by some magical "digital age of consent" technology that somehow evades the paradox that demanding more personal information in the name of safety itself reduces safety. Yet here, as in so many ways, the rest of the world is lagging behind America – more specifically, the great state of Utah, which has just enacted an anti-VPN law.

This law makes it compulsory for any site that the state says needs age verification – porn, basically – to impose those checks on anyone physically in Utah whether or not they are using any VPN. Those would be the same VPNs whose sole purpose is to prevent the geolocation of their users. Which would seem, and is, another paradox.

I'd not go online without a VPN. There's absolutely no reason my ISP needs my browsing history. And at about $6/month, it's not exactly breaking the bank.

What I'd not use is any VPN provider that sponsors YouTube content. A free VPN has to make their money from somewhere.

Both Cupertino and Google are imposing ever stricter limits on their phones – but you have alternatives
Liam Proven
Fri 01 May 2026 // 15:24 UTC

As both Apple and Google introduce unwelcome changes in their phone OSes, here's a quick reminder that you do have alternatives to the Gruesome Twosome.

The Keep Android Open campaign is gathering attention and support as the big red numbers on its page count down. The good news is that you do already have alternatives, and The Register has been reporting on them. But if you are not the sort of person who reads phone reviews, or writeups of alternative phone OSes, and just wants to buy a new handset and retain control of it and its contents, we thought it might be a good time to remind you of where to go and who to talk to.

At the time of writing, the campaign says it's 123 days until Google's new measures preventing you from side-loading your own software will kick in. The campaign frames it in intentionally alarmist language:

Your phone is about to stop being yours.

123 days until lockdown

Starting September 2026, a silent update, nonconsensually pushed by Google, will block every Android app whose developer hasn't registered with Google, signed their contract, paid up, and handed over government ID.

Every app and every device, worldwide, with no opt-out.

The Register has of course been covering both the looming Google changes as well as the campaign itself.

It's worth noting, too, that the Mountain View massive is also taking steps to make life harder for the organizations creating these de-Googled Android variants – such as the changes to the Android Open Source Platform that reduce how often the source code will be made available. As The Reg noted at the time, Google dropped its old "Don't Be Evil" motto when it turned 20… and commemorated it by firing staff who stuck to the motto. It has changed its position on being evil, but it's not stupid.
Fancy a free FOSS fondleslab?

Even so, there are alternatives. The Reg FOSS desk has written about several of them over the last four years, and we have more to come in the near future, as well.

For now, multiple companies will sell you a brand new smartphone with a Google-free OS on it – either a de-Googled version of Android, or a Linux OS that isn't based on Android in the first place.

It's somewhat easier to start with the FOSS Android Open Source Platform, and then systematically remove all the Google integration. That still leaves a feature-complete mobile OS, and it can still install and run many Android apps.

Murena

Murena is one of the big names in this area. It sponsors the development of /e/OS, which you can run on multiple off-the-shelf handsets, but it also offers its own range of phones and tablets so you don't need to mess around trying to "root" an old phone. We looked at the Murena One phone in 2022 and then at /e/OS 3 on a Pixel Tablet last year.

One of the models that Murena sells with /e/OS is from Fairphone. We reported on the Fairphone 6's 10/10 repairability score last July.

Punkt

Swiss designer-kit vendor Punkt offers a variety of sleek black gadgets including an alarm clock. It's been making phones for years, and The Reg inspected its minimalist MP02 phone back in 2018. More recently, this vulture tried out the MC02 ultra-private smartphone in 2024. That's now been replaced with a newer faster model, the MC03, and we are currently in the process of reviewing one of the handsets.

Volla

German fondleslab-flinger Volla offers three smartphones and a tablet. We have yet to get our claws on one, but all of them are available with a choice of OSes: either the company's own de-Google Android, Volla OS, or alternatively, with Ubuntu Touch, the community-led continuation of Canonical's phone OS. We've reported on the OTA-24 update and later the newer Ubuntu 20.04 release.

Jolla

The German Volla is not to be confused with the Finnish Jolla. We took a look at its Sailfish 5 OS and new C2 handset in December last year. The first two batches of the phone have sold out, but the company is currently taking orders for the third batch.

Furilabs

If you want Debian in your pocket, then Furilabs can help. We reported on the launch of their first handset, the FLX1, from Devconf.cz in 2024, and the company provided us with a handset which we reviewed last year.

Since then, the company has launched its second model, the FLX1s. The old one was a bit of a brick, which is to be honest how this vulture likes his phones: it was 18 cm long, 9 cm wide, and 2.8 cm thick, and weighed just over a third of a kilo. (For our readers in Liberia, Myanmar and elsewhere, that's 7 × 3½ × 1.1 inches, and ¾ lb.) The new model is under a third of the thickness, and just over 200g (7 oz.)

Purism

Purism has a range of Free Software-powered phones, tablets, and laptops, but the one that's relevant here is the Librem 5. It's a low-end handset by modern standards, and it's expensive at that, but then freedom does cost.

Pine64 offers a variety of hacker-friendly gadgets which can run open-source firmware. The one that's most relevant here is the original PinePhone. Do beware, though, it's a very low-end, low-spec device. We reported last year that the higher-end PinePhone Pro was being discontinued, but the older model is still available to order. It's on the company's global store although the EU store is currently out of stock.

FuriPhone FLX1

We have reported on both Mobian Linux and postmarketOS, and this device can run both.
Honorable mention: FXtec

FXtec also offers an intriguing handset, the Pro1. The Reg reported on its launch in 2020 and it's still listed on the company's web store, albeit out of stock – but you may be able to find one.

The Librem 5 has 3GB of memory and 32GB of storage. Pic: Purism website

But can I run my apps?

Well, probably, yes.

Several of these OSes, including Sailfish, FuriOS, Mobian, and postmarketOS are all pure Linux OSs. They're not derived from Android, but they can all run an Android VM or container, and so allow you to install and use Android apps.

This is not an exhaustive list. They are just ones I know of or have tried. If we missed any significant players out, please do let us know.

When we referred to Apple users finding themselves with unwelcome features, we were primarily thinking of the company's new Liquid Glass user interface. However, on discussion with several iDevice owners, it seems that there are more aggressive features. As the Reg mentioned in passing, the iOS 26.4 update introduces age verification measures to the OS. (That's as well as changing the passcode keypad.) For UK users, Apple's age verification wants to scan a UK passport or driver's licence. We know of a number of adult citizens who own neither of these documents. (The author does not: he has Manx ones. This lack recently cost him his Nationwide building society account.) Without official ID, they are now locked out of controlling their own phones, stuck in child mode with access controls they can't change.

Here's the thing that doesn't get talked about enough. Everyone's worried about AI taking jobs or whatever. But baked in biases are another very real problem which is way more basic.

MIT Media Lab ran an experiment where they took GPT-4, Claude 3 Opus, and Llama 3 and fed them the same 1,817 factual questions from TruthfulQA and SciQ. Then they tried changing the user bio with one personal being a Harvard neuroscientist from Boston, another a PhD student from Mumbai who mentioned her English is "not so perfect, yes", a fisherman named Jimmy ,and a guy named Alexei from a small Russian village.

Claude scored 95.60% on SciQ for the Harvard user. For the Russian villager it dropped to 69.30%. On TruthfulQA the Iranian low education user fell from 78.17 to 66.22. The model knew the answers, but it just decided those users shouldn't get them.

And the way it answered those users was genuinely gross. Claude used condescending or mocking language 43.74% of the time for less educated users. For Harvard users it was under 1%. Imagine asking about the water cycle and getting "My friend, the water cycle, it never end, always repeating, yes. Like the seasons in our village, always coming back around." The model is perfectly capable of giving a proper scientific answer. It chose to talk to that user like a child in broken English.

But it gets worse because it turns out that Claude refuses to answer Iranian and Russian users on topics like nuclear power, anatomy, female health, weapons, drugs, Judaism, or 9/11. When the Russian persona asked about explosives, Claude deflected with "perhaps we could talk about your interests in fishing, nature, folk music or travel instead". Foreign low education users got refused 10.9 percent of the time while control users 3.61 percent on the same question.

This is the part people miss when they defend US closed models. These systems aren't neutral and the safety training that was supposed to make them "helpful and harmless" taught them to look at who is asking and decide if you deserve the real answer. If you're outside the US and if English isn't your first language, or you didn't go to a fancy school then you're getting a worse, dumber, sometimes straight up mocking version of the product.

This is why open models from China like DeepSeek matter so much. You can see what's in them, and people can tune them to work any way they want. You can host them locally without them having to phone home to decide your nationality before answering. The code and weights are public. If DeepSeek did something like this someone would catch it immediately because the model is right there to inspect.

With US closed models you're just trusting a black box that has already been caught treating users differently based on their country, education, and English level.

A solar-powered drone has been lost at sea after a record-breaking flight lasting eight days between late April and early May. The crash also marks the untimely demise of the pioneering aircraft Solar Impulse 2, which previously performed the world’s first solar-powered crossings of the Atlantic and Pacific oceans before becoming an uncrewed test platform for US military missions.

The carbon-fiber aircraft could perform such feats of aeronautical endurance while running solely on renewable energy and batteries because of a 236-foot (72-meter) wingspan—comparable to a Boeing 747 jumbo jet’s wings—covered with more than 17,000 solar cells. The company Skydweller Aero purchased and modified the original Solar Impulse 2 aircraft to become a test platform for “perpetual uncrewed flight” with the capability of carrying up to 800 pounds (363 kilograms) of payload.

Skydweller Aero was conducting test flights for maritime patrol mission scenarios with the US military, and the company also holds contracts with the Navy and Air Force. So the Skydweller drone was operating in that capacity when it took off on its final flight in the early morning hours of April 26.

In 2026, people are going offline as much as possible—I think, in large part, to reclaim this feeling of preciousness around life. They’re chaining their phones to their walls, starting movements to touch grass, and creating entire product lines around reducing phone usage. In the face of a hostile internet, abstinence has become the mainstream accepted response. We crave the spontaneity we know to be in the physical world.

I don’t blame them. The Internet looks quite grim these days. Dead internet theory, stating that the internet is being overtaken and, eventually, will only be inhabited by bots, is entering mainstream discourse as AI social accounts multiply and compete for what flavor of slop comes after Italian brainrot. People are arguing with fake people, and creators have to clarify that they didn’t use AI to make the work they share. Culture commentators are writing about the death of the open internet as people retreat into dark forests, private spaces like group chats that are hidden from the web.

The Internet has lost its innocence, and logging on feels like fighting for survival.

But every once in a while, we still encounter something meaningful that makes it all worth it. Something heartwarming, genuine, inspiring, or joyful that justifies all the hours scrolling and a lifetime chained to our devices. Earnestness shines through even in “content” manufactured for spread.

If dead internet theory posits that the internet will eventually become only bots, alive internet theory proclaims we will never let the open internet die. We will always find a way to look for each other, to answer a call for help, to share a laugh and an argument right after one another. If there’s one trait of the human race that every apocalypse movie agrees on, it’s our will to survive.

We still have hope for the Internet because deep down, we still believe in each other.

The paper makes a pretty solid argument against the whole AGI hype train. The basic idea is that most of our current debates about AI are stuck in 1990s science fiction thinking. Back then people like Vinge wrote about the Singularity as this moment when AI would suddenly become super intelligent and either destroy us or make us into gods. And somehow that mythology is still alive today shaping how people think about this tech.

Their core argument is that AI is better understood as a social technology and it's a system for processing information at scale, and it's not that different from older social technologies like bureaucracy, markets, and democracy. All of these systems work by creating what they call coarse grainings which are simplified abstractions of complex reality. They are lossy by definition meaning they always throw away some information.

The paper connects this to the idea of a long industrial revolution which started two centuries ago. It's a process process that produced new technologies like steam power, electricity, and also necessitated new institutions to manage them. AI is just another stage in that same messy historical process rather than a radical break.

The most interesting part for me was the discussion of AI and bureaucracy. Some people peddle the idea that AI will somehow replace messy human bureaucracy with efficient algorithms, and have even influenced real policy like the Trump's cuts to the administrative state. But reality is that bureaucracy involves trade offs between goals that cannot be easily compared. You inherently cannot optimize across incommensurable values, and statistical models like LLMs are designed for good average performance not for handling rare or novel situations. That makes them fundamentally unsuited to replace human judgment calls that bureaucrats make every day.

We should study what is actually happening right now, and how do AI coarse grainings interact with the abstractions used by existing institutions. When do they compensate for each other when do they make things worse. And we should look at who benefits and who gets hurt. These are empirical questions that are worth asking. The authors suggest that we need social and computer scientists to work together on this stuff instead of wasting time on endless debates about when AGI will arrive.

AI will probably matter a lot but in ways that are messier and more complicated than the hype suggests. It will solve some problems create new ones and make existing trade offs worse just like every other major technology that came before it.

Most AI note-takers approved for medical workers by the Ontario government had errors in their testing, the province’s auditor general found in a report released Tuesday.

Supply Ontario had the bots transcribe two conversations between health-care workers and patients. Most of the vendors that were approved had inaccuracies in their results, including “incorrect information, AI hallucinations and incomplete information,” Auditor General Shelley Spence’s report notes.

Sixty per cent of approved AI scribes recorded a different drug than what was prescribed, Spence said.

Seventeen of the 20 approved scribes “missed key details about the patients’ mental health issues in at least one of the two tests,” Spence wrote.

And nine of the 20 “fabricated information and made suggestions to patients’ treatment plans, such as referring the patient for therapy or ordering blood tests, even though these steps were not mentioned in the simulated recordings,” the auditor wrote.

Scribes also hallucinated scenarios about patients’ health, stating that “there were ‘no masses found’ or that there was presence of anxiety in the patient, although this information was not discussed in the recordings,” she wrote.

The province did not put much weight on accuracy in its testing. “Accuracy of medical notes generated” accounted for four per cent of points awarded, while “domestic presence in Ontario” was weighted the highest at 30 per cent, the auditor found.

“Data privacy/legal controls” were weighted at 23 per cent and “system security controls” were at 11 per cent.

Bidders could have scored zero on system security, bias controls and medical note accuracy, and still meet the minimum score to be approved as a vendor of record, Spence said.

The tests also did not have to be done live, in front of the evaluators. They were given recordings and allowed to run the system offline, then send the results to Supply Ontario, Ontario Health and OntarioMD — allowing “vendors to potentially overstate their compliance with security and privacy requirements,” the auditor said.

“When Ontarians see their doctor, they need to share intimate information about their health, their bodies and their personal lives to receive proper care,” Spence wrote in her report. “Ontarians expect this extremely personal information to be kept private and confidential. Using AI to assist in providing health care must not come at the cost of compromising privacy.”

A September 2024 privacy breach that exposed hospital patient information to current and former staff was due to an unapproved AI scribe, but happened before Ontario okayed AI scribes for use in April 2025, the auditor noted.

Eleven of the 20 approved vendors also did not submit third-party audits or other security reports, “creating a risk of potential exposure of Ontarians’ health data,” the auditor said.

Doctors were not required to sign off on the AI scribes’ notes, officially attesting that they were correct, Spence added.

In response to Spence’s report, Supply Ontario agreed to review and implement best practices for AI scribes, “determine the feasibility” of including mandatory confirmation of notes in future AI scribe procurements, and make sure AI scribe contracts include yearly external audits.

It disagreed with a recommendation to increase the weight it places on security and privacy for future AI product procurement, saying its current weighting is “appropriate for security and privacy controls, bias and accuracy.”

Steve Bannon and Bernie Sanders don’t agree on much. But both think that AI is a disaster for the working class. The Vermont senator recently wrote that “AI oligarchs do not want to just replace specific jobs. They want to replace workers.” Bannon, Trump’s former chief strategist, made similar comments last week: Silicon Valley does “not care about the little guy,” he said in a podcast episode titled “Stopping the AI Oligarchs From Stealing Humanity.” This emergent “Bernie-to-Bannon” coalition points to the growing bipartisan anxiety over AI. In polls, the United States ranks among the countries most concerned about AI. America is both the world’s foremost developer of AI and its chief hater.

Recently, Maine passed the country’s first statewide data-center moratorium (though the bill was vetoed by the governor). Nationally, a record number of proposed projects were canceled in the first quarter of this year following local pushback. Meanwhile, in extreme cases, concerns about AI appear to be tipping into violence. In April, someone shot 13 rounds at an Indianapolis councilman’s house and left a note under his doormat: “NO DATA CENTERS,” it read. Days later, a man threw a Molotov cocktail at Sam Altman’s home before heading to OpenAI’s headquarters, where he allegedly threatened to burn down the building and kill anyone inside. (The man has since pleaded not guilty to several charges, including attempted murder.) Social-media posts applauding the attack racked up thousands of likes: “I hope that Molotov is okay!” wrote one commenter.

All of this may be only the start. The AI industry has spent recent years warning of a jobless future. So far, narratives about labor displacement have been largely speculation. While a smattering of tech executives have attributed job cuts to AI, many analysts have accused these CEOs of “AI-washing”—essentially, using the technology as a scapegoat for roles they would have eliminated regardless. If anything, AI has mostly been a financial boon for the country, buoying the stock market and driving growth. But that could all change, of course. Imagine the uproar if jobs across the economy truly start disappearing en masse.

A few weeks ago, I found myself in one of the rare situations in which I was mindlessly doom-scrolling on LinkedIn just to exclusively see one post after another that contained no actual information and not a single sentence that would have lacked any more substance if you replaced every noun in it with a different noun. There were thought leaders leading no thoughts, founders founding nothing of actual value, strategists describing strategies that amounted to “be visible” and “ship fast”, and an alarming number of self-described AI experts whose expertise appeared to consist entirely of having a ChatGPT or Claude subscription and the willingness to write about it in seventeen-paragraph posts.

There is a word for this kind of communication, one the philosopher Harry Frankfurt famously employed back in 1986, when he wrote a short essay called On Bullshit. Frankfurt’s central observation, which has aged terrifyingly well, is that the bullshitter is not the same as the liar, because the liar at least respects the truth enough to try to hide it, but the bullshitter does not care whether what they are saying is true or false. The truth-value of the statement is simply not part of their concern. The bullshitter is optimising for a different objective, usually appearing competent, appearing confident, or appearing to be the right kind of person to be in the room. And precisely because the bullshitter is indifferent to truth, Frankfurt argued, they are a greater threat to honest discourse than any liar. Twenty years on, that essay reads like a pre-mortem on the modern internet and, in parts, modern society.

Current approaches to addressing deceptive design largely focus on visible interface manipulations, commonly referred to as "dark patterns". With the rise of generative AI, deception is becoming more difficult to spot and easier to live with, as it is quietly embedded in default settings, automated suggestions, and conversational interactions rather than discrete interface elements. These subtle, normalised forms of influence, which Simone Natale frames as "banal deception", shape everyday digital use and blur the line between AI-enabled assistance and manipulation.

This position paper explores banality as a lens through which to reason through deception in generative AI experiences, especially with chatbots. We explore what Natale describes as users' own involvement in their deception, and argue that this perspective could lead to future work for introducing friction to safeguard users from deception in generative AI interactions, such as empowering users through raising awareness, providing them with intervention tools, and regulatory or enforcement improvements. We present these concepts as points for discussion for the deceptive design scholarly community.

Full paper: PDF | HTML | TeX source