The French government has collected nearly €10 million in additional taxes after using machine learning to spot undeclared swimming pools in aerial photos. In France, housing taxes are calculated based on a property’s rental value, so homeowners who don’t declare swimming pools are potentially avoiding hundreds of euros in additional payments.

The project to spot the undeclared pools began last October, with IT firm Capgemini working with Google to analyze publicly available aerial photos taken by France’s National Institute of Geographic and Forest Information. Software was developed to identify pools, with this information then cross-referenced with national tax and property registries.

The project is somewhat limited in scope, and has so far analyzed photos covering only nine of France’s 96 metropolitan departments. But even in these areas, officials discovered 20,356 undeclared pools, according to an announcement this week from France’s tax office, the General Directorate of Public Finance (DGFiP), first reported by Le Parisien.

As of 2020, it was estimated that France had around 3.2 million private swimming pools, but constructions have reportedly surged as more people worked from home during COVID-19 lockdowns, and summer temperatures have soared across Europe.

Ownership of private pools has become somewhat contentious in France this year, as the country has suffered from a historic drought that has emptied rivers of water. An MP for the French Green party (Europe Écologie les Verts) made headlines after refusing to rule out a ban on the construction of new private pools. The MP, Julien Bayou, said such a ban could be used as a “last resort” response. He later clarified his remarks on Twitter, saying: “[T]here are ALREADY restrictions on water use, for washing cars and sometimes for filling swimming pools. The challenge is not to ban swimming pools, it is to guarantee our vital water needs.”

France’s tax offices, the DGFiP (known more commonly as Le Fisc), says it now plans to expand the use of its AI-pool-spotter to the entirety of metropolitan France (excluding the country’s overseas departments), which could net an additional €40 million in taxes.

Early reports on the project suggested that the machine learning software had an unusually high error rate of 30 percent, and regularly mistook other architectural features — like solar panel installations — for swimming pools. Now, though, Le Fisc says it’s ironed out these problems, and is looking to expand the use of its software spotting pools to identifying other undeclared and taxable housing improvements, like extensions and annexes.

“We are particularly targeting house extensions like verandas, but we have to be sure that the software can find buildings with a large footprint and not the dog kennel or the children’s playhouse,” Antoine Magnant, the deputy director general of public finances, told Le Parisien, reports The Guardian.

The power demands of data centers can be staggering, with larger facilities using as much electricity as tens of thousands of homes. Now, in parts of London, those energy requirements are reportedly stopping developers from building new houses because there’s simply not enough electricity to go around.

According to a report from The Financial Times, which shared letters sent to housing developers by London’s local governing body (the Greater London Authority, or GLA), the issue is affecting new developments in three west London boroughs: Ealing, Hillingdon, and Hounslow. The GLA said west London’s electricity grid was at capacity, and told one developer there may not be “sufficient electrical capacity for a new connection” until 2035.

It seems there are a few factors at play here, including the difficulty of building new infrastructure given the high value of land in London. But a significant one is the electricity demands of data centers located along the M4 corridor. This is a stretch of land adjacent to the M4 motorway that runs west of London and is home to many technology companies, including campuses for Microsoft, Oracle, LG, Huawei, Amazon, and Dell.

The GLA notes in its letter that “data canters use large quantities of electricity, the equivalent of towns or small cities, to power servers and ensure resilience in service.” It’s not clear how many housing projects have been blocked by this bottleneck, but the FT notes that the three London boroughs affected account for around 11 percent of the city’s housing supply.

The electricity demands of data centers are only likely to grow in future, as companies switch more processing and storage operations to cloud operations, and as rising temperatures caused by climate change strain facilities’ cooling systems. Earlier this month, for example, when the UK experienced a record-breaking heatwave, the rise in temperature knocked out several data centers in London operated by Oracle and Google. It also led to energy companies paying record prices to keep the lights on: a worrisome combination.

The UK has been suffering from a chronic under-supply of housing for many years now, with the problem regularly described as one of the country’s greatest challenges.

A survey of news consumption habits has found that TikTok is the fastest growing news source among adults in the UK. According to a report from UK regulator Ofcom, the app is currently used by 7 percent of UK adults as a news source, compared to 1 percent in 2020.

The survey is more evidence of TikTok’s massive and growing presence in the digital world, and illustrates how the app is quickly becoming a go-to source for information as well as entertainment. In the US, for example, it was recently estimated that 40 percent of Gen Z (individuals born 1997 to 2012) use either TikTok or Instagram for everyday searches to find things like where to eat, instead of previous favorites like Google Search and Maps.

Although TikTok has risen fast as a news source in the UK, it’s still only the sixth most popular destination among those aged 16 to 24 (where it’s used by 27 percent of the demographic). That’s behind the BBC website/app (29 percent), Twitter (35 percent), news TV channel BBC One (36 percent), Facebook (40 percent), and Instagram (46 percent).

Among 12 to 15 year olds, the skew towards social media news is much greater. The top three digital news sources for this group are Instagram (29 percent), then YouTube and TikTok (both at 28 percent). However, these news sources are still outpaced by the two most important for teens: talking to family (65 percent) and watching TV (59 percent).

“Teenagers today are increasingly unlikely to pick up a newspaper or tune into TV news, instead preferring to keep up-to-date by scrolling through their social feeds,” said Yih-Choung Teh, Ofcom’s group director for strategy and research in a press statement. “And while youngsters find news on social media to be less reliable, they rate these services more highly for serving up a range of opinions on the day’s topical stories.”

This may be a worrying trend for some given TikTok’s reputation as a source of disinformation. (Recent analysis of news related to the Ukraine-Russia war, for example, found that TikTok was a prime vector for spreading “false narratives” about the conflict.) But users seem to be somewhat aware of this problem, with Ofcom noting that social media platforms continue to score worse than rival news sources on attributes like “trust,” with around two-thirds of users of social media sites saying they don’t trust them for news.

There’s been another twist in the saga of Meta’s attempt to buy GIF-sharing service Giphy, Bloomberg reports. Judges with the Competition Appeal Tribunal have told the UK’s competition regulator, the Competition and Markets Authority (CMA), to reconsider its previous ruling telling Meta to sell Giphy. “We have agreed to reconsider our decision in light of this finding,” the CMA told Bloomberg in a statement. It hopes the process will take no longer than three months.

The ruling comes a month after the same tribunal largely sided with the regulator after Meta appealed its order to unwind the deal and sell Giphy. The tribunal found in favor of the CMA on five out of the six claims, but, crucially, on the remaining claim, it ruled that the CMA had failed to properly inform Meta of Snapchat’s acquisition of Gfycat, thus undermining the company’s defense. Now, to remediate the error, the CMA will reconsider its ruling, and Meta will have the opportunity to comment on the CMA’s final report.

Meta (then known as Facebook) announced its intention to buy Giphy over two years ago, in May 2020. But the deal quickly attracted the attention of the UK’s competition regulator, who was concerned that, among other things, Meta’s ownership of the popular gif-sharing service could limit other platforms’ access to its content. As a result, last November, the CMA officially ordered Meta to unwind the deal and sell Giphy

A representative from Meta declined to comment to The Verge on the latest ruling. But in response to the tribunal’s findings last month, Meta spokesperson Christopher Sgro reiterated that “We firmly believe our investment would enhance Giphy’s product for the millions of people, businesses, and partners who use it.”

The heads of the FBI and UK domestic security service shared a platform for the first time to issue dire warnings about the threats posed by the Chinese government’s espionage operations, BBC News reports. FBI Director Christopher Wray and MI5 Director General Ken McCallum were speaking at a joint event at MI5’s London headquarters in front of an audience that included business CEOs and senior figures from universities.

“The Chinese government is set on stealing your technology—whatever it is that makes your industry tick—and using it to undercut your business and dominate your market,” the Wall Street Journal reports Wray said in the speech. The FBI director added that the benefits of keeping a piece of technology confidential may sometimes outweigh those of accessing the Chinese market.

“Maintaining a technological edge may do more to increase a company’s value than would partnering with a Chinese company to sell into that huge Chinese market, only to find the Chinese government and your partner stealing and copying your innovation,” Wray said, adding that it represents “an even more serious threat to western businesses than even many sophisticated businesspeople realized.”

In their speech, the two allege that the Chinese government is engaged in a “coordinated campaign” to gain access to important technology, and to “cheat and steal on a massive scale.” They added that the Chinese government’s hacking program dwarfs that of every major country, and that it has a global network of intelligence operatives. The threat means that MI5 is running seven times as many investigations into Chinese activity as it was four years ago, while the FBI is opening roughly two new counterintelligence investigations every day, The Wall Street Journal reports.

“Today is the first time the heads of the FBI and MI5 have shared a public platform,” MI5’s McCallum said. “We’re doing so to send the clearest signal we can on a massive shared challenge: China.” He added that the threat is “real and it’s pressing” and that it could be “the most game-changing challenge we face.”

In terms of specific examples, MI5’s McCallum cited the case of a British aviation expert who was offered a job by a company that was actually a front for Chinese intelligence officers looking to acquire technical information on military aircraft, BBC News reports. Another engineering firm came close to making a deal with a Chinese company, before seeing its technology taken and the deal called off. The incident forced the company into bankruptcy.

A spokesperson for China’s embassy in Washington, Liu Pengyu, denied the allegations, telling the Associated Press in a statement that the country’s government “firmly opposes and combats all forms of cyber-attacks” and “will never encourage, support or condone” them. The Chinese government maintains that it does not interfere in the affairs of other countries, but that it will defend itself against cyberattacks. The statement criticized “U.S. politicians who have been tarnishing China’s image and painting China as a threat with false accusations,” the Wall Street Journal reports.

Amazon has agreed to simplify the process of canceling Prime in Europe, meaning customers in the region will be able to end their subscription in just two clicks, the European Commission has announced. The changes, which were implemented as of July 1st, should bring to an end the “multiple pages” filled with “distracting information” and “unclear button labels” that Amazon has previously used to add friction to the cancellation process.

The European Commission says the changes apply to the European Union and European Economic Area. Despite the UK having left the former as of the beginning of 2020, The Guardian reports that UK subscribers will also benefit from the two-click unsubscribe process. But when contacted for comment, a spokesperson for Amazon dodged questions about whether it would implement similar changes in the US, saying it has “no changes to announce at this time.”

Image: European Commission

“Customer transparency and trust are top priorities for us,” spokesperson Bradley Mattinger said in an emailed statement. “By design we make it clear and simple for customers to both sign up for or cancel their Prime membership. We continually listen to feedback and look for ways to improve the customer experience, and have no changes to announce at this time.”

Amazon has agreed to make the changes following a complaint by EU consumer groups including the Norwegian Consumer Council, which produced a lengthy report on Amazon’s opaque Prime cancellation process in January 2021. The report contains screenshots of the multiple pages users have to scroll through to cancel a subscription, which it said contains “manipulative design techniques” also known as “dark patterns.”

Screenshots: Amazon

The European Commission says Amazon needed to change its process to comply with the bloc’s Unfair commercial practices directive. Amazon had previously agreed to change its web interface to label the cancel button more clearly and shorten distracting text, but now it says it’ll shorten this explanatory text even further. The changes will be made across desktop, mobile, and tablet.

“Consumers must be able to exercise their rights without any pressure from platforms,” Commissioner for Justice Didier Reynders said in a statement. “One thing is clear: manipulative design or “dark patterns” must be banned. I welcome Amazon’s commitment to simplify their practices to allow consumers to unsubscribe freely and easily.”

Future EU legislation could clamp down on user interfaces like these even further. The forthcoming Digital Services Act (DSA), which EU lawmakers provisionally agreed to earlier this year is expected to contain explicit prohibitions related to the use of “dark patterns.” The DSA is expected to come into force 15 months after being voted into law, or from January 1st, 2024, whichever is later.

Samsung has started producing 3nm chips, beating rival chip manufacturer TSMC to the more power efficient fabrication process, Bloomberg reports. TSMC’s 3nm process isn’t expected to go into mass production until the second half of 2022.

Samsung says the new fabrication process is 45 percent more power efficient than its previous 5nm process, has 23 percent higher performance, and 16 percent smaller surface area. In the future, it hopes its second generation 3nm process can reduce power consumption and size by 50 percent and 35 percent respectively, and increase performance by 30 percent.

The announcement is a key milestone in Samsung’s efforts to compete with TSMC, which dominates the market for contract chip production and is the manufacturer of Apple’s chips for its iPhones, iPads, MacBooks, and Macs. But Bloomberg reports that Samsung is unlikely to be able to make inroads against TSMC’s market share until it can prove the cost-efficiency of its new 3nm process is competitive with the market leader.

Samsung says the chips will initially be produced for “high performance, low power computing” applications, but that it plans for them to eventually make their way to mobile. The chips will be manufactured in South Korea for now, initially at its Hwaseong facilities before expanding to Pyeongtaek, Bloomberg notes. The company’s forthcoming chip plant in Texas, which early reports said could eventually have the capability to produce 3nm chips, is not scheduled to start mass manufacturing until 2024.

The location of chip manufacturing has come under a huge amount of scrutiny as a result of the recent global chip shortage, with the US, Europe, and China vying to have manufacturers produce more chips domestically.

Grocery app Gorillas, which promises to deliver goods in as quickly as 10 minutes, is laying off half of its office staff. In a press release, the company said it was letting go of roughly 300 employees from a “global office workforce” of 600. (This workforce also includes roughly 14,400 staff working in warehouse and as delivery drivers.)

The company is also planning to tighten its focus on five markets that account for 90 percent of its revenue: the UK, US, Germany, France, and the Netherlands. The company also operates in four other European markets — Spain, Denmark, Italy, and Belgium — where it says it is “looking at all possible strategic options for the Gorillas brand.” That might mean pulling out of these markets, but Gorillas tells The Verge nothing has been decided yet.

The news suggests trouble for the fast-growing “instant” grocery sector. Over the past few years, a huge number of these services have sprung up around Europe, fueled by venture capital investment and pandemic lock-down orders. These companies all rely on the same basic infrastructure: warehouses filled with groceries scattered throughout urban centers and armies of e-bike and scooter-riding delivery drivers to deliver the goods.

Many experts have warned that these services are essentially unsustainable, though, with their growth predicated on discounted prices that have been underwritten by VC funds hoping to capture what remains of the market once a few dominant players emerge. A report by The Information last August suggested that one upcoming player, JOKR, was losing as much as $159 on each order it delivered in the US.

Some countries have also objected to the impact these firms have on urban infrastructure. In the Netherlands, for example, grocery apps including Flink, Getir, and Gorillas have been forced to limit the speed of delivery drivers’ e-bikes and keep their warehouses out of certain built-up areas — both measures a response to complaints from city residents.

Combined with the global downturn affecting the tech sector more broadly, it seems some delivery apps that have struggled to grow may be in trouble. Sifted reports that the company had trouble raising more funding, and that in May of this year, the Gorillas app was downloaded just 320,000 times — compared to 1.5 million downloads for Turkish rival Getir.

When Facebook briefly blocked all news from being posted on its platform in Australia last year, it used an overly broad definition of news publisher that it knew would cause collateral damage, company whistleblowers are alleging. Complaints filed with Australia’s Competition & Consumer Commission and the US Department of Justice allege that the company engaged in “a criminal conspiracy to obtain a thing of value, namely favorable regulatory treatment,” The Wall Street Journal reports.

The news ban was put into place by Facebook last February in protest over a proposed Australian law that would effectively force platform-holders like it and Google to pay news publishers when sharing their content. But the ban was chaotically implemented, and there were widespread reports of it blocking government organizations and nonprofits alongside news publishers.

Blocked non-media pages included government organizations like the Department of Fire and Emergency Services Western Australia and Queensland Health and nonprofits like Mission Australia and the Hobart Woman’s shelter. The ban took place during fire season in Australia, the WSJ notes, and coincided with the country’s COVID-19 vaccine rollout. In total, the WSJ reports that Facebook internally recognized it had blocked 17,000 pages that it shouldn’t have in the first day of the ban.

“It was clear this was not us complying with the law, but a hit on civic institutions and emergency services in Australia,” one whistleblower said, in comments reported by the WSJ. Facebook started removing the pages on February 18th, after the House of Representatives passed an initial version of the bill, but ahead of final votes by it and the Australian Senate the following week.

Facebook publicly admitted at the time that it had taken a “broad definition” of news content, in light of what it said was a lack of “clear guidance” in the law. Records of internal conversations obtained by the WSJ offer more evidence that this was the case. “[The proposed Australian law] we are responding to is extremely broad, so guidance from the policy and legal team has been to be overinclusive and refine as we get more information,” wrote a product manager in an internal log.

Leaked documents suggest the company classified a page a news publisher if over 60 percent of the content it shared was classified as news. Documents also suggest that the company had planned on excluding all government and education domains from the ban.

But the list of organizations who saw their Facebook pages removed as a result of the ban suggests these safeguards malfunctioned, and according to the WSJ, Facebook ignored approaches that could have more precisely targeted news organizations. The company decided against using its database of news publishers known as the News Page Index, apparently because it relied on news publishers to opt-in, and the complaints allege it failed to use existing whitelisting tools to protect important accounts. It failed to put an appeals process in place before blocking pages, and did not notify affected pages in advance.

Despite the technical issues with the media ban, Facebook officials internally praised the company’s response to the legislation. Facebook’s head of partnerships Campbell Brown called the team’s efforts “genius,” while chief operating officer Sheryl Sandberg said “the thoughtfulness of the strategy, precision of execution, and ability to stay nimble as things evolved [set] a new high-standard.” CEO Mark Zuckerberg praised the team’s ability to “execute quickly and take a principled approach.”

When reached for comment, Facebook spokesperson Gina Murphy sent over a statement saying “the documents in question clearly show that we intended to exempt Australian government Pages from restrictions in an effort to minimize the impact of this misguided and harmful legislation. When we were unable to do so as intended due to a technical error, we apologized and worked to correct it. Any suggestion to the contrary is categorically and obviously false.”

The law Facebook was protesting went on to pass later that month. But it did so while containing amended language that means the Australian Treasurer needs to consider private deals struck between publishers and platforms before designating a company like Facebook as a platform under the law, and allowing it to be forced into arbitration. As a result, over a year after the law passed, neither Facebook nor Google have officially been designated as platforms under the rules.

Six months after predicting the global chip shortage would last until at least 2023, Intel CEO Pat Gelsinger is now suggesting it might be 2024 before we’re fully out of the woods.

“[W]e believe the overall semiconductor shortage will now drift into 2024, from our earlier estimates in 2023, just because the shortages have now hit equipment and some of those factory ramps will be more challenged,” he told CNBC on Friday.

But while that sounds a little doom-y and gloom-y, you should know that the “chip shortage” is an involved, evolving situation that doesn’t affect every kind of chip at every time. Some industries and some kinds of parts have been worse hit than others as things go on. In fact, Intel’s own chips are doing fairly well. “For the first time in years, Intel fabs and our substrate supply are close to meeting our customers’ demand,” Gelsinger said yesterday on the company’s Q1 2022 earnings call.

When Gelsinger says that the shortage will stretch into 2024, he’s partly talking about the industry’s ability to meet demand for new products being built on new lines, not just existing ones. “We expect the industry will continue to see challenges until at least 2024 in areas like foundry capacity and tool availability as an IDM,” he said on yesterday’s call. Digitimes recently reported that chipmaking equipment suppliers are now backed up more than 18 months, up from six months just last year.

Put another way, CPUs, GPUs, and game consoles were some of the highest-profile items hit by shortages, but it seems like supply and demand are already beginning to equalize there. But networking chip vendors are still in the middle of a significant chip shortage: Gelsinger namedropped ethernet as a particularly difficult “ecosystem supply constraint” that has slowed down PC shipments.

But that’s not why Intel’s Client Computing Group (which handles consumer processors, among other things) is down 13 percent this quarter. Intel attributed that to a “ramp-down of the Apple CPU and modem business” and “OEM inventory burn” as well as “lower consumer and education demand” — aka schools are buying fewer Chromebooks and Apple has all but fully transitioned away from Intel to its own M1 processor, which left Intel laptops in the dust.

Intel is one of the companies investing heavily in new production lines, by the way, building new fabs in Ohio, Arizona, and Germany, though the current timeline suggests none of those new fabs will go online until the chip shortage is over. The first new fabs in Chandler, Arizona, aren’t scheduled to open until 2024.