Michael Zhang writes via PetaPixel: In a post titled "High Fidelity Image Generation Using Diffusion Models" published on the Google AI Blog (and spotted by DPR), Google researchers in the company's Brain Team share about new breakthroughs they've made in image super-resolution. [...] The first approach is called SR3, or Super-Resolution via Repeated Refinement. Here's the technical explanation: "SR3 is a super-resolution diffusion model that takes as input a low-resolution image, and builds a corresponding high resolution image from pure noise," Google writes. "The model is trained on an image corruption process in which noise is progressively added to a high-resolution image until only pure noise remains." "It then learns to reverse this process, beginning from pure noise and progressively removing noise to reach a target distribution through the guidance of the input low-resolution image." SR3 has been found to work well on upscaling portraits and natural images. When used to do 8x upscaling on faces, it has a "confusion rate" of nearly 50% while existing methods only go up to 34%, suggesting that the results are indeed photo-realistic.

Once Google saw how effective SR3 was in upscaling photos, the company went a step further with a second approach called CDM, a class-conditional diffusion model. "CDM is a class-conditional diffusion model trained on ImageNet data to generate high-resolution natural images," Google writes. "Since ImageNet is a difficult, high-entropy dataset, we built CDM as a cascade of multiple diffusion models. This cascade approach involves chaining together multiple generative models over several spatial resolutions: one diffusion model that generates data at a low resolution, followed by a sequence of SR3 super-resolution diffusion models that gradually increase the resolution of the generated image to the highest resolution." "With SR3 and CDM, we have pushed the performance of diffusion models to state-of-the-art on super-resolution and class-conditional ImageNet generation benchmarks," Google researchers write. "We are excited to further test the limits of diffusion models for a wide variety of generative modeling problems."

According to a study from Finbold, consumers spent an estimated $41.5 billion on apps from the Apple App Store during the first six months of 2021, which is almost double compared to $23.4 billion spent by Android consumers. 9to5Mac reports: The Apple App Store spendings in H1 21 represents a growth of 22.05% from a similar period in 2020. Cumulatively, spending on the two platforms recorded a growth of 24.8%, year-over-year hitting a total of $64.9 billion: "Mobile app spending grew in 2021 mainly due to consumer behavior resulting from the containment measures around the coronavirus pandemic. From the H1 spending, the outstanding assumption is that consumers' demand might not be slowing down, considering that most global jurisdictions have eased the pandemic control measures with the vaccine rollout."

The report shows that although the App Store spending has remained significant, Google Play appears to be catching up by recording the highest growth rate between H1 2020 and H1 2021: 30% YoY. So even with the App Store doing way better, "Android flourished as the pandemic remained persistent in countries with an Android-dense population." The Finbold survey shows that to keep on the top, Apple's are countering with the Small Business Program, which reduces the App Store commission from 30% to 15% for developers grossing less than $1 million a year. Although the report shows that the spending on the App Store and Google Play Store were mainly dominated by the gaming sector, in which consumers globally spend $10.32 billion on mobile games in the first half of 2021, the survey highlights that the top three grossing mobile apps worldwide for H1 2021 are TikTok, YouTube, and Tinder.

General Motors is halting production of the Chevy Bolt EV after a number of battery fires prompted a massive recall that has already cost the company nearly $2 billion. From a report: The automaker won't resume making the 2022 Bolt and Bolt EUV until at least mid-September. It had initially stopped production last week because of the global chip shortage. The recall repair process is also on hold, as GM is still waiting to get new battery modules from its supplier, South Korean conglomerate LG, that it is confident are free of defects. "We will not resume repairs or restart production until we are confident LG is producing defect free products for us," GM spokesman Daniel Flores said in a statement to The Verge.

Popular instant messaging app Telegram has joined the elite club of apps that have been downloaded over 1 billion times globally, according to Sensor Tower. From a report: The Dubai-headquartered app, which was launched in late 2013, surpassed the milestone on Friday, the mobile insight firm told TechCrunch. As is the case with the app's chief rival, WhatsApp, India is the largest market for Telegram. The world's second largest internet market represents approximately 22% of its lifetime installs, Sensor Tower said.

China kicked off a two-month campaign to crack down on commercial platforms and social media accounts that post finance-related information that's deemed harmful to its economy. From a report: The initiative will focus on rectifying violations including those that "maliciously" bad-mouth China's financial markets and falsely interpret domestic policies and economic data, the Cyberspace Administration of China said in a statement late Friday. Those who republish foreign media reports or commentaries that falsely interpret domestic financial topics "without taking a stance or making a judgment" will also be targeted, it added. The move is aimed at cultivating a "benign" online environment for public opinion that can facilitate "sustainable and healthy development" of China's economy and its society, according to the statement. It followed a draft proposal issued earlier Friday by the cyberspace regulator to regulate algorithms that technology firms use to recommend videos and other content. Commercial websites and platforms will be ordered to clean up financial information posts and shut accounts deemed in violation, under the supervision of authorities including the cyberspace administrator, the finance ministry, central bank as well as securities, banking and insurance regulators.

Alphabet's Google generated $11.2 billion in revenue from its mobile app store in 2019, according to a court filing unsealed on Saturday, offering a clear view into the service's financial results for the first time. From a report: Attorneys general for Utah and 36 other U.S. states or districts suing Google over alleged antitrust violations with the app store also said in the newly unredacted filing that the business in 2019 had $8.5 billion in gross profit and $7 billion in operating income, for an operating margin of over 62%. The figures include sales of apps, in-app purchase and app store ads. Google told Reuters the data "are being used to mischaracterize our business in a meritless lawsuit."

The company and its accusers said in a separate filing on Saturday a trial in late 2022 is possible over whether Google abuses its alleged monopoly in app sales for Android devices. In its quarterly financial disclosures, Google groups Play app revenue with that of other services and accounts for the store's ad revenue as part of another broader category. Attorneys general, as well as mobile app developer Epic Games and others separately suing Google, have contended that it generates huge profits through the Play Store by taking 30% of the fee for every digital good sold inside an app. The plaintiffs say Google's cut is arbitrarily high, siphoning app developers' profits.

Although Apple is making some changes to its App Store policies and setting up a $100 million fund for small developers, critics say that the terms of a class-action settlement don't meaningfully loosen the company's grip on its digital marketplace. From a report: Spotify Chief Legal Officer Horacio Gutierrez: "Apple's proposed concessions fail to address the most basic aspects of their anticompetitive and unfair App Store practices. They are attempting to distract policymakers and regulators and slow down the momentum that's building around the world to address their behavior."

Match Group: "This is a raw demonstration of their monopolistic power: making capricious changes designed to spur good PR for their benefit right as legislation, regulatory scrutiny and developer complaints are closing in on them. We hope everyone sees this for what it is -- a sham."

Sen. Amy Klobuchar (D-NY): "This new action by Apple is a good first step towards addressing some of these competition concerns, but more must be done to ensure an open, competitive mobile app marketplace, including commonsense legislation to set rules of the road for dominant app stores."

A new article at ZDNet argues that "The 1990s had a word for being trapped inside a manipulative notion of human contact: AOL."

"Facebook and its ilk are the rebirth of that limited vision." Once upon a time, roughly thirty years ago, there was a computer network called America Online... There was already an Internet, but most people didn't know how to use it or even that it existed. AOL, and a couple of competitors, Compuserve and Prodigy, offered people online things they could do, such as chat with other people... The services had only one drawback, which was that they were limited. People couldn't do just whatever they wanted, they could only pick from a small menu of functions, such as chat, that the services provided... As it grew and grew, the World Wide Web became an amazing place in contrast to AOL... People were so excited by the World Wide Web, they never wanted to go back to AOL or Compuserve or Prodigy. The three services withered...

People got excited about Facebook because it was a place where they could find real people they knew, just like MySpace, but also because it had some features like AOL, like the game Farmville. Business people were even more excited because Facebook started to generate a lot of advertising revenue. Advertisers liked Facebook because it not only knew who was talking to whom, it also knew a little bit about the hobbies and interests of people. Advertisers liked that because they could use the information to "target" their ads like never before. Smart people said that Facebook had what are known as "network effects." It became more powerful the more people joined it...

There were just a couple problems with Facebook. Facebook was a lot like AOL. It limited people by telling them with whom they could communicate.... One of the bad things was that people no longer had control. They had given so much information about themselves to Facebook and its competitors that it was like those companies owned people when they were in Cyberspace. The services didn't seem to do a great job of handling people's information, either.
What's interesting about this article is it even tells you how the story ends: Then one day, someone smart built a new technology that didn't require people to sign away their information. Now, people could meet anyone they wanted and talk about whatever they wanted, not just what Facebook or its competitors said was okay. People felt more relaxed, too, because even though there were ads, people could meet up in Cyberspace without every single action they took being used to fuel an advertising machine.

People got excited again, like the first time they found the Web and gave up on AOL.

But there our story ends, because that chapter has not yet been written.

A columnist for Inc. writes that Google "may reduce the salaries of employees who choose to work at home full-time, based on the cost of living where they live, according to an internal calculator viewed by Reuters."

They also argue that Google's move is "likely to be a disaster." It may seem sensible, given that a salary that barely covers a San Francisco studio apartment might get you a mansion in, say, Topeka. That's the logic Google says it's using. "Our compensation packages have always been determined by location," a spokesperson told Reuters.

But cutting pay for existing employees who opt to work from home is a terrible idea and it shows a complete lack of emotional intelligence. If Google is smart, it will shelve this idea. So will Facebook, Twitter, the UK government, and any other company considering a similar move. Here's why:

1. A salary is about more than just paying the bills... In real life, a pay cut will feel like an insult to most employees, even if it has nothing to do with their performance or their value to the company. You're literally telling them that they're worth less. Is that the message you want them to hear?

2. Google is being greedy... Like other tech giants, it's thrived during the pandemic. Cutting people's salaries when your share price has more than doubled, your revenues are up 62 percent, and your profits are up even more seems like the pinnacle of corporate greed. Not a good look.

3. It will make Google even more unequal than it already is...

Megan Borovicka forget all about her Facebook account after 2013, reports the Washington Post. "But Facebook never forgot about her." The 42-year-old Oakland, California, lawyer never picked any "friends," posted any status updates, liked any photos or even opened the Facebook app on her phone. Yet over the last decade, Facebook has used an invisible data vacuum to suction up very specific details about her life — from her brand of underwear to where she received her paycheck... It isn't just the Facebook app that's gobbling up your information. Facebook is so big, it has convinced millions of other businesses, apps and websites to also snoop on its behalf. Even when you're not actively using Facebook. Even when you're not online. Even, perhaps, if you've never had a Facebook account.

Here's how it works: Facebook provides its business partners tracking software they embed in apps, websites and loyalty programs. Any business or group that needs to do digital advertising has little choice but to feed your activities into Facebook's vacuum: your grocer, politicians and, yes, even the paywall page for this newspaper's website. Behind the scenes, Facebook takes in this data and tries to match it up to your account. It sits under your name in a part of your profile your friends can't see, but Facebook uses to shape your experience online. Among the 100 most popular smartphone apps, you can find Facebook software in 61 of them, app research firm Sensor Tower told me. Facebook also has trackers in about 25 percent of websites, according to privacy software maker Ghostery...

Facebook got a notice when I opened Hulu to watch TV. Facebook knew when I went shopping for paint, a rocking chair and fancy beans. Facebook learned I read the websites What To Expect, Lullaby Trust and Happiest Baby. Over two weeks, Facebook tracked me on at least 95 different apps, websites and businesses, and those are just the ones I know about. It was as if Facebook had hired a private eye to prepare a dossier about my life. Why does Facebook think that's okay? The company emailed me answers about how its tracking technology works, but declined my requests to interview its chief privacy officer or other executives about its alleged monopoly....

Who in their right mind thought they were signing up for this much surveillance back when they first joined Facebook?
The article points out that in 2014 Facebook began allowing its advertisers to target users based on websites they'd visited...and now also gathers more data about users from other companies. And "While many companies were using browser cookies, which could be easily cleared or blocked, Facebook tied what it learned to real identities — the names on our Facebook profiles." And beyond that, companies "can report other identifying information to Facebook like your email to help it figure out who you are... If you've never had a Facebook account at all? It may still be watching."

It's a lucrative business, the Post points out. "In 2013, the average American's data was worth about $19 per year in advertising sales to Facebook, according to its financial statements. In 2020, your data was worth $164 per year."

What does Facebook know about your off-Facebook activity? You can find out at this URL.

If you just want to stop them from giving this information to advertisers, the right side of that page has an option to "Clear History — Disconnect off-Facebook activity history from your account." But you then have to also click "More Options" and then "Manage Future Activity" to also stop them from later matching up more of your off-Facebook activity to your profile for advertisers.

If you try to select it, Facebook warns what you'll be missing — that "Keeping your future off-Facebook activity saved with your account allows us to personalize your experience." And proceeding anyways then generates a popup reminding you that "We'll still receive activity from the businesses and organizations you visit. It may be used for measurement purposes and to make improvements to our ads systems, but it will be disconnected from your account."

And apparently your activity on Oculus isn't covered, and will still remain connected to your Facebook account.

"Toyota has apologised for the 'overconfidence' of a self-driving bus," reports the Guardian — after the slow-moving bus hit a Paralympic judo expert. Toyota added that it would temporarily suspend the service, with Toyota's president saying the event "shows that autonomous vehicles are not yet realistic for normal roads." The Japanese athlete, Aramitsu Kitazono, will be unable to compete in his 81kg category this weekend after being left with cuts and bruises following the impact with the "e-Palette" vehicle...

As part of its sponsorship of Tokyo 2020, Toyota has been showcasing its autonomous vehicles via a shuttle service, which has been running around the clock in the athletes' village. On Thursday, however, one of the buses pulled away from a T-junction and drove through a pedestrian crossing while Kitazono, a visually impaired athlete, was walking across.

Tokyo police said that vehicle operators had told them they "were aware that a person was there but thought [the person] would [realize that a bus was coming] and stop crossing the [street]", according to the Asahi Shimbun newspaper.
CNN cites reports that the vehicle was under manual control at the time of the accident, adding that the vehicle "was barely moving, but it still managed to collide with a visually-impaired athlete at the Paralympic Games, raising potential concerns about the limitations of autonomous driving technology."

Infosecurity magazine reports: Some of the world's biggest tech companies have committed tens of billions of dollars to improving supply chain security, closing industry skills gaps and driving security awareness among the public, according to the White House.

As reported by Infosecurity yesterday, the Biden administration welcomed the CEOs of Microsoft, Apple, Google, IBM and others to a meeting yesterday to discuss the "whole-of-nation" effort needed to address cybersecurity threats." The result of that encounter has been a series of commitments from these firms, including $10bn from Google over the next five years to expand zero trust and improve supply chain and open source security. The tech giant will apparently also help 100,000 Americans earn "digital skills certificates."

IBM said it would train 150,000 people in cyber skills over the coming three years and focus on improving the diversity of the security workforce, while Microsoft has committed $20bn over five years to drive security by design, and $150m for federal, local and state governments. Apple will establish a new program to improve supply chain security, including among its 9000 US suppliers, with multi-factor authentication (MFA), vulnerability remediation, event logging and incident response all playing a key role. Amazon is making MFA devices available to all AWS customers and rolling out the security training it offers employees to the general public.

Aside from these commitments, the White House announced the expansion of its Industrial Control Systems Cybersecurity Initiative, from the electricity sector to natural gas pipelines, and said the National Institute of Standards and Technology (NIST) would develop a new framework for supply chain security. In another potentially significant move, insurer Resilience said it would require policyholders to meet a threshold of cybersecurity best practice as a condition of receiving coverage — something experts have been demanding for some time across the industry.
NextGov.com also quotes the president's remarks about a cybersecurity executive order issued May 12th: "Because of that order, government will only buy tech products that meet certain cybersecurity standards, which will have a ripple effect across the software industry, in our view, ultimately improving security for all Americans,"

The New York Times tells the story of an Indiana physician spreading misinformation about the pandemic. Public health officials say statements like his have contributed to America's vaccine hesitancy and resistance to mask-wearing, exacerbating the pandemic. His videos "have amassed nearly 100 million likes and shares on Facebook, 6.2 million views on Twitter, at least 2.8 million views on YouTube and over 940,000 video views on Instagram." His talk's popularity points to one of the more striking paradoxes of the pandemic. Even as many doctors fight to save the lives of people sick with Covid-19, a tiny number of their medical peers have had an outsize influence at propelling false and misleading information about the virus and vaccines.

Now there is a growing call among medical groups to discipline physicians spreading incorrect information. The Federation of State Medical Boards, which represents the groups that license and discipline doctors, recommended last month that states consider action against doctors who share false medical claims, including suspending or revoking medical licenses. The American Medical Association says spreading misinformation violates the code of ethics that licensed doctors agree to follow.

"When a doctor speaks, people pay attention," said Dr. Humayun Chaudhry, president of the Federation of State Medical Boards. "The title of being a physician lends credibility to what people say to the general public. That's why it is so important that these doctors don't spread misinformation."

In Facebook's "widely viewed content report" released last week, The Verge's Casey Newton noticed something arguably just as damning as the spread of COVID-19 misinformation or rise of vaccine hesitancy: almost all of the most-viewed posts on Facebook over the past quarter were effectively plagiarized from elsewhere. From the report: Facebook's report details the top 20 most widely viewed posts on the network over the past three months. One of the posts was deleted before Facebook published it. Of the remaining 19, though, only four appear to have been original. The remaining 15 had been published in at least one other place first, and were then re-uploaded to Facebook, sometimes with small changes. [...] Facebook has long been home to reappropriated content, from the freebooting scandal during 2017's pivot to video to the more recent phenomenon of Instagram's Reels being flooded with videos bearing TikTok watermarks. But this kind of dumb, cheap growth hacking should sound familiar to anyone who paid even passing attention to the 2016 election. Russia's infamous Internet Research Agency commissioned a troll army to build up big followings on innocuous-seeming Facebook pages using a wide variety of engagement bait, then gradually shifted those pages to begin sharing more divisive political memes.

That's all much harder to do now, thanks to a variety of measures Facebook has taken to make it more difficult for people to disguise their identities or countries of origin. The company now routinely removes networks of pages where the creators' identities are suspect. And it's worth saying that in the most recent election, inauthentic behavior of the 2016 variety did not play a significant role. Most importantly, Facebook now has a policy against "abusive audience building" -- switching topics and repeatedly changing a page's name for the purpose of growing a following. But it seems notable that for domestic actors, the tactics not only work, but remain the most effective way to reach a large audience five years later. Steal some questions that went viral somewhere else, spam them on your page, and presto: you're one of the most-viewed links for the entire quarter on the world's biggest social network. "The plagiarists who dominate Facebook's top 20 links are likely doing it primarily for clout and ill-gotten audience growth," Casey goes on to say. "But some of the other characters here appear to have more direct monetary incentives..."

Just months after a CEO shakeup, Waymo is officially halting sales of its custom sensors to third parties. TechCrunch reports: The move sees the Alphabet-owned self-driving company unwinding a business operation just two years into its lifespan. Waymo confirmed the decision to Reuters, adding that it's now focusing on deploying its Waymo Driver tech across its Waymo One ride-hailing and Waymo Via trucking divisions. [...] Waymo began selling LiDARs -- the tech that measures distance with pulses of laser light -- to companies barring its autonomous vehicle rivals in 2019. It initially planned to sell its short-range sensor (known as Laser Bear Honeycomb) to businesses in the robotics, security and agricultural technology sectors. A form on its website also lists drones, mapping and entertainment as applicable industries.

Waymo's fifth-generation Driver technology uses an array of sensors -- including radar, lidar, and cameras -- to help its cars "see" 360 degrees during the day and night, and even in tough weather conditions such as rain or fog. While its simulated and real world driving tests have helped it to amass a massive dataset that is crunched using machine learning-based software. According to anonymous sources cited by Reuters, Waymo intends to use in-house tech and external suppliers for its next-gen LiDARs.

T-Mobile said a cyberattack earlier this month that exposed millions of customer records was carried out using specialized tools to gain entry to the network, followed by brute force-style hacking techniques to access user data. From a report: "In short, this individual's intent was to break in and steal data, and they succeeded," Chief Executive Officer Mike Sievert said Friday in a statement, the company's fullest account yet of what happened. The company has hired cybersecurity provider Mandiant and consulting firm KPMG to improve its defenses, he said. The breach, the fourth that has compromised T-Mobile customer records in as many years, involved personal information including names, dates of birth, Social Security numbers and driver's license information. Sievert said the company is working with law enforcement and can't share further details of what happened. Further reading: T-Mobile Hacker Explains How He Breached Carrier's Security.

An anonymous reader quotes a report from TechCrunch: China is not done with curbing the influence local internet services have assumed in the world's most populous market. Following a widening series of regulatory crackdowns in recent months, the nation on Friday issued draft guidelines on regulating the algorithms firms run to make recommendations to users. In a 30-point draft guideline published on Friday, the Cyberspace Administration of China (CAC) proposed forbidding companies from deploying algorithms that "encourage addiction or high consumption" and endanger national security or disrupt the public order.

The services must abide by business ethics and principles of fairness and their algorithms must not be used to create fake user accounts or create other false impressions, said the guidelines from the internet watchdog, which reports to a central leadership group chaired by President Xi Jinping. The watchdog said it will be taking public feedback on the new guidelines for a month (until September 26). The guidelines also propose that users should be provided with the ability to easily turn off algorithm recommendations. Algorithm providers who have the power to influence public opinion or mobilize the citizens need to get an approval from the CAC. Friday's guidelines appear to target ByteDance, Alibaba Group, Tencent and Didi and other companies whose services are built on top of proprietary algorithms.

Unnervingly futuristic, these bulky, high-tech toothbrushes promise to scrub your choppers thoroughly in 20 seconds. WSJ: Dentists like Dr. Lana Rozenberg are overly familiar with two eternal fibs: that their clients floss regularly, and that they brush their teeth for at least two minutes twice a day. "Most people don't brush their teeth for two minutes," said the Manhattan-based industry veteran. "Thirty seconds is more like it," or under a second for each of their 32 teeth. But what if, in those 30 seconds, a device could reach the front, back and sides of every tooth at once? That's the proposition of new "whole mouth" toothbrushes, which rely on vibration and a preponderance of bristles packed inside a structure resembling a mouthguard to deliver an up-to-snuff scrubbing in as little as 20 seconds -- 10 each for top and bottom sets of teeth.

"It helps make things way faster, way easier and feels a little more guaranteed because you actually feel it on each one of your teeth," said Kristopher Paul, a medical-transportation driver in St. Petersburg, Fla., who has bit down on a 360 Sonic Brush Pro ($70) each morning for the past year. Mr. Paul, 36, also likes the tool's 15-minute whitening mode, which combines a blue LED light and whitening gel to fade bothersome coffee stains.

Elise Blanchard, writing on Mozilla blog: [...]

What happened in 1993 to suddenly make hyperlinks blue? No one knows, but I have some theories. I often hear that blue was chosen as the hyperlink color for color contrast. Well, even though the W3C wasn't created until 1994, and so the standards for which we judge web accessibility weren't yet defined, if we look at the contrast between black as a text color, and blue as a link color, there is a contrast ratio of 2.3:1, which would not pass as enough color contrast between the blue hyperlink and the black text. Instead, I like to imagine that Cello and Mosaic were both inspired by the same trends happening in user interface design at the time. My theory is that Windows 3.1 had just come out a few months before the beginning of both projects, and this interface was the first to use blue prominently as a selection color, paving the way for blue to be used as a hyperlink color.

Additionally, we know that Mosaic was inspired by ViolaWWW, and kept the same gray background and black text that they used for their interface. Reviewing Mosaic's release notes, we see in release 0.7 black text with underlines appearing as the preferred way of conveying hyperlinks, and we can infer that was still the case until something happened around mid April right before when blue hyperlinks made their appearance in release 0.13. In fact, conveying links as black text with underlines had been the standard since 1985 with Microsoft 1, which some once claimed Microsoft had stolen from Apple's Lisa's look and feel.

I think the real reason why we have blue hyperlinks is simply because color monitors were becoming more popular around this time. Mosaic as a product also became popular, and blue hyperlinks went along for the ride. Mosaic came out during an important time where support for color monitors was shifting; the standard was for hyperlinks to use black text with some sort of underline, hover state or border. Mosaic chose to use blue, and they chose to port their browser for multiple operating systems. This helped Mosaic become the standard browser for internet use, and helped solidify its user interface as the default language for interacting with the web.

Microsoft is announcing today that it won't block people from installing Windows 11 on most older PCs. While the software maker has recommended hardware requirements for Windows 11 -- which it's largely sticking to -- a restriction to install the OS will only be enforced when you try to upgrade from Windows 10 to Windows 11 through Windows Update. From a report: This means anyone with a PC with an older CPU that doesn't officially pass the upgrade test can still go ahead and download an ISO file of Windows 11 and install the OS manually. Microsoft announced its Windows 11 minimum hardware requirements in June, and made it clear that only Intel 8th Gen and beyond CPUs were officially supported. Microsoft now tells us that this install workaround is designed primarily for businesses to evaluate Windows 11, and that people can upgrade at their own risk as the company can't guarantee driver compatibility and overall system reliability. Microsoft won't be recommending or advertising this method of installing Windows 11 to consumers.