Tech

How a new data center could impact your town’s electric bills and environment

Published

51 minutes ago

18 May 2026

NewsAdmin

The only good data center is a canceled data center.

Or so a growing number of Americans seem to feel.

Throughout the United States, citizens are mobilizing against the construction of new data centers in general — and the massive, “hyperscale” ones that fuel artificial intelligence, in particular.

Data centers can increase local air pollution, though their emissions vary widely between different contexts. Their impact on local water supplies, meanwhile, has been greatly exaggerated.
Data centers often drive up an area’s electricity bills.
But hyperscale campuses can deliver major economic benefits to their host communities, including job growth and tax revenue.
In some cases, the benefits of data center development almost certainly outweigh the costs.

The past four years have witnessed an unprecedented boom in the construction of such facilities, driven by AI firms’ insatiable thirst for computing power. Between 2022 and 2025, annual spending on the creation of data centers in the United States jumped from $15 billion to over $35 billion, in constant dollars.

And many Americans have had enough.

At the local level, municipalities are nixing data center projects at a historic clip. Over the first three months of this year, plans for at least 20 such facilities have been shelved amid public backlash, according to an analysis by Heatmap Pro. Together, those canceled projects represent $41.7 billion in forgone investment.

In statehouses and Congress, meanwhile, lawmakers are pushing to freeze data center construction outright. Last month, Maine’s state legislature passed a moratorium on new data centers in the state. Gov. Janet Mills ultimately vetoed that legislation, but insisted that she too supported freezing all data center construction in Maine, except for a single, long-planned project in the economically challenged town of Jay.

At least 12 other states are entertaining data center moratoria, while four municipalities have imposed permanent bans. In the Senate, Bernie Sanders has introduced a bill that would pause the construction of AI supercomputing campuses nationwide, until a long list of regulations and social programs are enacted.

This rebellion against data centers is partly motivated by concerns about artificial intelligence. Some progressives, right-wing populists, and tech-wary centrists believe that unfettered AI development poses intolerably high risks — to workers’ economic security, the Earth’s climate, and/or humanity’s survival. From this perspective, the point of blocking new data centers is primarily to throw sand into the gears of AI progress. And if artificial intelligence really is on the cusp of wrecking civilization, then trying to choke off Big Tech’s access to computing power makes some sense.

That said, much of the resistance to data centers is rooted in anxieties about the buildings themselves.

Many Americans have come to think that these industrial complexes offer little to their host communities beyond economic burdens and ecological devastation. Judging by activist rhetoric and viral media accounts, data centers invariably slurp up localities’ water, pollute their air, despoil their landscapes, and poison their residents with “infrasounds” — all while driving up municipalities’ electricity bills and sponging off their tax dollars.

But the truth is more complicated.

If a large data center comes to your town, it could make the local environment slightly worse — and your electric bill somewhat higher. But it could also raise your municipality’s wages, while reducing its property tax rates.

The scale of these costs and benefits vary widely from place to place. Yet in areas with sound environmental regulations, relatively clean and robust electric grids, and progressive tax codes, data centers tend to be a “win-win” for both residents and large tech companies (assuming the latter don’t get the former killed in a robot apocalypse, anyway).

So, how do you decide whether your community should welcome hundreds of acres of computing hardware? Let’s examine the (real and imagined) environmental harms and material upsides point by point — and how the balance between them shifts with local conditions.

Data centers can be dirty

Data centers do come with some environmental costs. But the scale and uniformity of their local impacts are often exaggerated.

The most serious of these harms is air pollution.

A hyperscale campus can require as much electricity annually as a midsize American city, as such complexes must power tens of thousands of continuously running processors — and then dissipate the consequent heat with industrial-scale cooling systems.

In some places, data centers meet these gargantuan energy demands by spitting a ton of particulates into the sky. For example, xAI’s Colossus campus in Memphis is partly powered by 35 on-site natural gas turbines. All that combustion appears to have dramatically increased the concentration of nitrogen dioxide in nearby air, with peak levels jumping 79 percent since the facility opened in 2024, according to researchers at the University of Tennessee, Knoxville.

Elon Musk’s Tennessee operation is exceptionally dirty. But more typical data centers also tend to increase regional air pollution, at least marginally, by boosting the utilization of an area’s natural gas or coal plants and running backup diesel generators when their energy needs temporarily exceed the grid’s supply.

Nevertheless, in all these cases, the problem arises from the type of energy that a data center uses, not from some inextricable feature of all server farms.

Data centers sited in regions with abundant non-carbon energy sources produce relatively little air pollution, at least by the standards of industrial enterprises. According to Google’s official data, its Oregon cloud computing operations run exclusively on non-carbon energy sources about 87 percent of the time.

Water is less of a problem

On most other fronts, data centers’ environmental costs tend to be overstated.

Server farms do need substantial amounts of water. But this is true of almost all forms of industrial production. And modern data centers are not exceptionally water-intensive operations, in part because they typically use closed-loop cooling systems that recirculate the same pool of H2O repeatedly.

According to the calculations of AI researcher Andy Masley, as of 2023, 0.04 percent of America’s fresh water was being consumed inside data centers. For comparison, the nation’s golf courses churned through 33 times as much H2O that year.

To be sure, data centers don’t need to consume a significant share of the national water supply to burden especially arid regions. Fortunately, parched localities tend to already constrain the amount of local water available for industrial purposes.

Consider Box Elder County, Utah. There, investors are seeking to build a massive hyperscale campus over sagebrush in the state’s hinterlands. This has prompted furious pushback from local residents, driven partly by fears that the facility would deplete the rapidly declining Great Salt Lake. If that body of water continues losing volume, it will release plumes of toxic dust over Utah’s urban core.

In this context, it’s understandable that residents would fear a vast data center soaking up their region’s scarce reserves. And yet, to secure its H2O, the Box Elder project needed to purchase water rights from an existing agricultural user. This transaction did not increase the total amount of industrial water consumption in Utah, but merely transferred a small portion of a fixed pool from one business to another.

Critically, as City Journal’s Shawn Regan notes, that exchange is plausibly net-positive for the Great Salt Lake: When farmers use water to irrigate a field, much of it gets lost to evapotranspiration and never returns back to its source. By contrast, when water courses through a closed-loop data center, it retains far more of its volume, and then gets periodically flushed back into the local watershed. Therefore, shifting water rights from a farm to a data center conceivably reduces long-term depletion.

Finally, when sited too close to residential areas, some data centers generate noise pollution that meaningfully degrades their neighbors’ quality of life. But server farms located in far-flung, industrial zones do not have this problem. And contrary to some viral videos, there is no reason to think that data centers emit subaudible “infrasounds” that damage the health of those in their vicinity.

Data centers create more jobs than you might think

So, data centers come with some real environmental costs, which can be profound in certain circumstances.

It does not follow, however, that server farms are always bad for the localities that host them. An industrial enterprise can benefit a community, even if it imposes some ecological burdens. Or so most people — including most prominent data center skeptics — seem to believe.

After all, when a town loses a car plant or steel mill to offshoring, progressives and right-wing populists don’t typically cheer that municipality’s good fortune — even though such manufacturing facilities can be far worse for local air quality than a data center.

The key question, then, is how a given data center’s ecological harms stack up against its economic benefits. And the latter, like the former, vary widely from one situation to another.

Hyperscale campuses can enrich communities in two primary ways: by generating jobs and tax revenue.

Data center opponents often belittle their benefits to employment. And not without reason. Once built, a server farm is among the least labor-intensive industrial facilities one can find.

Nonetheless, data center projects can meaningfully benefit a region’s workers in both the short run (by creating a large number of temporary construction jobs) and in the longer term (by fostering a favorable ecosystem for tech sector development).

A recent study from the Brookings Institution spotlights the latter point. To gauge data centers’ impacts on employment, researchers examined labor market trends in 93 counties that welcomed their first data center between 2008 and 2024 — and 3,000 counties that never received one. After controlling for a variety of variables, the authors estimate that the arrival of a large data center raised private employment in the former counties by 4 to 5 percent over a five- or six-year period. Meanwhile, the data centers also appeared to lift wages for both existing workers and new hires by about 3 to 4 percent.

These gains are driven partly by temporary spikes in construction employment. But hyperscale data centers — the kind that powers AI — also yielded more durable jobs in the information sector. As the authors note, vast cloud computing campuses generate demand for local fiber installers, IT contractors, managed service providers, and other types of tech firms. And these businesses generally open near such campuses, so as to regularly service them.

This dynamic is most visible in counties that host data center clusters: When places welcomed at least four new data centers in the studied time period, their employment soared by 23 percent. By contrast, counties with a single, non-hyperscale data center saw boosts to construction employment, but little durable growth in IT jobs.

Servers will sometimes pay your taxes

But the biggest local benefit of data centers is typically fiscal.

Hyperscale campuses generate a lot of tax revenue without consuming much in the way of public services. In many jurisdictions, a data center doesn’t just need to pay taxes on its land or buildings, but also on its exceedingly valuable equipment — servers, networking gear, generators, etc. For this reason, such facilities can yield far more revenue per acre than a housing complex or office park.

What’s more, unlike the residents of a new subdivision, servers in a hyperscale facility do not send any children to a municipality’s public schools, drive around on its roads, crowd its parks, or use public transit.

For these reasons, data centers can sometimes deliver immense fiscal value. In Loudoun County, Virginia, they now provide nearly half of the county’s tax revenue — enough to cover all of its government’s functions beyond the school system. In other words, Loudoun County residents effectively get their local police departments, courts, parks, infrastructure maintenance, and many other social services provided free of charge — a fact that has allowed the county to slash its property tax rate by about 30 percent over a decade.

To be sure, northern Virginia’s experience is somewhat exceptional; Loudoun has been a major IT infrastructure hub for decades, and hosts 200 data facilities. Nevertheless, other jurisdictions enjoy similar fiscal benefits at a smaller scale.

Indeed, data center development can be especially valuable for economically declining rural towns. Such municipalities often struggle to attract highly profitable, labor-intensive industries, since they lack large numbers of knowledge workers. But a single data center’s direct labor needs are fairly light, even as its commercial value is high. They can therefore provide rural localities with a modicum of economic development and revenue, when few other enterprises will.

For precisely this reason, however, localities sometimes get caught in a race to the bottom to attract data center investment, showering exorbitantly wealthy tech companies in tax incentives, a phenomenon that has further embittered much of the public against hyperscale projects.

Fortunately, there’s reason to think that most communities could be driving a harder bargain, without chasing away investment. As Brookings notes, many of these tax breaks appear to be subsidizing projects that would happen even in their absence. When making siting decisions, hyperscalers tend to care more about the availability of power, land, and fiber infrastructure than fiscal favors.

When a data center comes to your town, it will probably — though not definitely — push up your electric bill

Of course, data centers also come with one infamous economic downside: Their gargantuan energy demands can push up nearby residents’ electricity bills.

The mechanism here is straightforward: If you abruptly add the equivalent of a midsize city to a region’s electric grid, power demand will probably rise faster than utilities can expand production. In the short term, this may force them to purchase more expensive electricity from wholesale markets. In the longer run, it can require them to finance new power plants and transmission infrastructure. In either case, ratepayers are liable to foot some of the bill.

Market analysts and economists have produced evidence that this is already happening in some markets.

Even on this front, however, context matters. Rising electricity demand does not automatically translate into higher prices. And data center development does not unfailingly correlate with electricity inflation.

Virginia is, again, a case in point. The Old Dominion saw a 14 percent increase in electricity demand between 2019 and 2024, due in part to data centers’ energy consumption, according to a study from the Lawrence Berkeley National Laboratory. And yet, over that same period, electricity rates in the state actually fell by 1 cent per kilowatt hour, in inflation-adjusted terms.

Trends in North Dakota were even more paradoxical. That state saw electricity demand skyrocket by almost 40 percent, amid a data center building boom. Nevertheless, its inflation-adjusted electricity prices fell by 3 cents per kilowatt hour.

These results are counterintuitive. But they are consistent with the (often odd) economics of power markets. Electric grids aren’t built to provide the exact amount of power a region requires. Rather, they generally maintain excess generation and transmission capacity.

For this reason, in systems with a lot of spare capacity, moderate increases in demand can lead to fuller utilization of existing resources, rather than major new capital expenses. In such circumstances, adding a data center to the grid may spread its fixed costs across a larger volume of electricity sales, lowering the amount that each individual household must contribute to sustain the system.

To be clear, more often than not, adding a bunch of hyperscale data centers to a region’s grid is likely to push up its electricity prices. But that result is not automatic. And governments can limit ratepayers’ burdens by forcing hyperscalers to meet the full costs of satisfying their power demands.

New data centers may end the world; but first, they’ll probably lower someone’s property taxes

Although this catalog of data centers’ effects may seem complicated, the upshot is simple: data center projects are neither intrinsically good nor bad for localities. Each offers a distinct mix of costs and benefits, depending on factors including how its operations are fueled, regulated, and taxed.

Put differently, policymakers have the power to make data center development pay off for more communities — by, among other things, decarbonizing electric grids, enforcing noise limits, and paring back tax breaks for Big Tech firms. Regardless, even under today’s flawed policy frameworks, ordinary people often benefit when a data center comes to town.

Hyperscalers might eventually trigger a robot apocalypse. But in the meantime, they’ll almost certainly make some municipalities a bit better off.

Source link

Tech

Google has sold so much TPU capacity that its own researchers are queueing for the rest

Published

3 minutes ago

18 May 2026

NewsAdmin

Alphabet runs the most enviable AI infrastructure stack in the industry. The success of the third-party deals with Anthropic and Meta is the reason internal access has become a competitive resource.

Google has spent a decade quietly building the most enviable position in AI infrastructure: a healthy cloud business, its own custom chips, and supply deals that make its TPUs the default alternative to Nvidia for major external customers.

The success of that strategy has produced an internal problem that the company did not anticipate.

Bloomberg’s Julia Love reported on Monday that Google’s own AI researchers, including teams inside Google DeepMind, are now jockeying for access to the computing resources their employer is also selling to Anthropic and Meta.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

The structural cause is straightforward. Google has agreed to invest up to $40bn in Anthropic on a deal that includes five gigawatts of TPU capacity over five years and access to up to one million seventh-generation Ironwood chips.

A separate, Broadcom-mediated supply line covers a further 3.5GW of TPU capacity for Anthropic from 2027, building on the 1GW the company is already receiving in 2026. Anthropic itself has publicly described the Google TPU stack as central to its training and serving roadmap.

Meta, the other commercial-scale TPU customer Bloomberg names, signed its own deal earlier this year. The capacity those commitments lock up is capacity not available to Google’s internal model teams without queueing.

DeepMind’s chief executive Demis Hassabis said earlier this year that the constraint cuts in two directions. Some of the bottleneck is hardware: ‘a few suppliers of a few key components’, as he put it, with high-bandwidth memory from Samsung, Micron and SK Hynix the most-cited choke point.

Some of it is research throughput, because, in Hassabis’s words, researchers ‘need a lot of chips to be able to experiment on new ideas at a big enough scale’. The hardware constraint is partly outside Google’s control. The internal-allocation constraint is not.

The arithmetic underneath this is large. Alphabet is on a guided capex range of $175bn-$185bn for 2026, inside a combined Big Tech AI infrastructure spend that has crossed $650bn this year. Google has, on its own commentary, been bringing well over a gigawatt of new AI compute capacity online in 2026.

The decade-long bet on TPUs is finally producing the kind of unit-economics advantage that lets the company sell its chips, host its competitors’ models, and run its own frontier research on the same fabric. The fabric is just no longer big enough for all three uses at once.

Bloomberg’s reporting names two specific signals of the tension. Researchers including Ioannis Antonoglou, a long-tenured DeepMind contributor, have departed for startup roles in the past 18 months, a pattern that has accelerated as compute access has become harder to secure inside Google.

Oren Etzioni, the former Allen Institute for AI chief executive cited in the piece, has framed the dynamic publicly as the predictable result of an internal market in which compute is rationed by managerial seniority rather than by the unit-cost economics that govern external customer contracts.

Google has spent the past 18 months in a delicate position: it needs its TPU programme to demonstrate volume traction with named external customers to validate the technology against Nvidia, while keeping enough internal capacity for Gemini training runs and DeepMind research.

The four-partner inference-chip supply chain with Broadcom, MediaTek and Marvell is a hedge designed to relieve the constraint by adding capacity downstream of TPU training. It has not yet shipped at the scale the demand curve requires.

Google did not dispute Bloomberg’s internal-allocation framing on the record; it pointed to its broader infrastructure investment posture and the fact that compute constraints are a category-wide condition rather than a Google-specific one.

That is true on the evidence: every major model provider is, on the cleanest reading of Q1 2026 earnings, compute-constrained relative to its own research aspirations.

What makes the Google version newsworthy is the side-by-side: the company has, at the same time, become its main competitors’ largest infrastructure supplier. If it can keep selling the asset and using it is the question the next several quarters will settle.

Source link

Tech

5 Harbor Freight Icon Tools Under $25 Users Say Are Worth Buying

Published

15 minutes ago

18 May 2026

NewsAdmin

Harbor Freight is known for a wide range of tools that span affordable budget options to high-end equipment, like socket sets, power tool kits, and portable table saws. One of the best Harbor Freight brands you can buy, Icon has built a steady reputation for being a brand that even professionals can trust. On the Harbor Freight website, there are several Icon collections that you can choose from, such as hand tools, shop tools, lighting, storage, and diagnostics. Icon hand tools also have the benefit of lifetime warranty coverage, which means you can easily return it to Harbor Freight stores for a replacement if it breaks.

While they’re not the cheapest out of the bunch, Icon tools are generally more affordable than other comparable brands like Snap-On. Harbor Freight also regularly lists price comparisons on its product listings, which can be helpful if you’re on the fence about saving or splurging. Everyone’s favorite magnetic tool holder mat is just one of the highly rated Icon tools you can snag for under $25. So, if you’re looking to get started with your Icon tool collection on a budget, here are some things that you only have to buy once and never have to pay for again.

Icon Professional Scraper

While you can use your hands to remove things like dirt, adhesives, or paint, a scraper can help do the job without getting gunk under your nails. In some cases, a 3D-printed scraper may be enough, but if you need something a little more durable, the Icon Professional Scraper should be on your list. Designed for different hard surfaces, Icon’s scraper can be used for tile, metal, and glass, including those in high locations due to its shaft. It boasts an extended reach that goes up to 12.64 inches. Weighing 0.43 lbs (or a little less than an iPhone 17 Pro), the handle also has ergonomic comfort grips. The professional scraper comes with 10 high-carbon steel replacement blades, which you can switch out with its thumbscrew.

One of the cheapest Icon tools out there, you can get it for $13.99 at Harbor Freight, wherein it enjoys a mostly positive 4.6-star average rating from more than 400 people. With a pretty high 94% recommendation rate, you’re likely going to be in good hands, with several verified buyers saying that it works great for scraping everything from stickers to old window tint. That said, there were a few unhappy customers who rated it a single star, citing durability. While it can take different kinds of blades (plastic or metal), some customers noted that the ones that come with the kit tend to break easily.

Icon 12 in. S-Jaw Quick-Adjust Pliers Wrench

Designed to work with irregular surfaces, the $19.99 Icon S-Jaw Quick-Adjust Pliers can handle jobs that ordinary pliers may struggle with. According to Icon, it is great for tasks in confined spaces and ideal for automotive works with hex, round, or square shapes. Several reviews note that it can fulfill functions related to plumbing, too. The Swedish pipe wrench has features like angled teeth, hardened jaws, and mechanisms against unthreading. With a 1-⅞-inch throat depth and 3-inch jaw opening, the 12-inch model is the smallest offer in the line-up. But if you need a larger size, Icon also offers other S-Jaw Quick-Adjust Pliers models: the 17-inch ($29.99) and 21-inch ($39.99).

All together, the three sizes have been rated an impressive 4.7 stars by over 310 Harbor Freight customers. As of May 2026, 94% of customers recommend it, while 260+ people have rated it 5 stars. Most people said that it wasn’t just great at gripping, but that it is versatile. While people did mention a bit of a learning curve, one user claimed that it “Made the impossible possible, like a cross between vice grips and a pipe wrench.” Around 3% of reviewers, who rated it a single star, did note issues with slipping, locking, and the quick release feature.

Icon SAE Color-Coded L-Shape Ball End Hex Key Set

Hex keys (or Allen wrenches) are some of the most versatile tools out there that you can use for everything from furniture assembly, outdoor grills, to regularly adjusting bicycle parts. It’s no wonder that Icon manufacturers two highly-rated options under $25. Priced at $24.99, the 13-piece ICON Color-Coded L-Shape Ball End Hex Key Set is made of corrosion-resistant, premium steel construction and has two features that set it apart from other sets: color coding and the ball end, which offers more flexibility with 20-degree angle entry.

Available in both SAE or Metric, the colored hex key set boasts an average rating of 4.8 stars from 290+ people. Apart from 96% of customers recommending it, the large majority of Harbor Freight reviewers also gave it a perfect rating. Unsurprisingly, many users praised its colors, which they claimed helped them easily find what they needed. Some customers even love it so much that they recommended purchasing both sets. Although, there were a few people who mentioned that the paint did have a tendency to wear off after consistent use.

But if you want to slash $3 from the price and don’t really care about colors, ICON also sells the all black 13-piece L-Shaped Ball End Hex Key Set, which retails for just under $22. Although it is less popular, it did receive an average rating of 4.9 stars from 180 Harbor Freight customers with a 98% recommendation rate.

Icon 16 oz. Soft Face Dead Blow Hammer

Whether you’re joining some wood, working on cars, or installing some tires, a soft face dead blow hammer, like the Icon 16 oz. Soft Face Dead Blow Hammer can be incredibly useful. Unlike regular hammers, it’s less likely to damage delicate surfaces and create permanent dents. It also helps reduce the likelihood of injury from bouncing back. On its longest side, this Icon Dead Blow Hammer measures 13.63 inches (or about a little more than a standard ruler). It has a soft grip and wide flared handle with a 4-inch width, wherein there is structural steel shank. Made of Polyurethane, the Icon Soft Face Dead Blow Hammer head has a 2-inch rubberized striking face that is resistant to chemical damage.

Alternatively, if you need bigger hammers for the job, Icon also sells three other models with heavier head weights: 24 oz. ($24.99), 32 oz. ($29.99), and 48 oz. ($34.99). Collectively, the Icon Soft Face Dead Blow Hammers hold consistently stellar reviews on the Harbor Freight website. Out of more than 270 customers, every single one said that they’d recommend it. With an average rating of 4.9 stars, a large majority (90.5%) even gave a perfect 5 stars and not a single review went below 3 stars. Some of the common praises include how they’re both comfortable and versatile. One person even notes “I think they’re the best product Icon sells.”

Icon Professional Mini Soft-Grip Pick and Hook Set

Sometimes, your fingernails need some extra help, so a good pick and hook set can save you a lot of hassle. Available in two colors (red and green), the Icon Professional Mini Soft-Grip Pick and Hook Set contains four hand tools: a hook, awl, 90-degree pick, and 45-degree pick. Measuring 6 inches in length with a 3-inch shaft, each tool has a soft grip and precision tips. Designed to withstand various chemicals in your job site, it comes with a storage tray.

Priced at $21.99, the Icon Professional Mini Soft-Grip Pick and Hook Set has been rated 4.7 stars on average by 680+ Harbor Freight customers. Among the 94% of users saying that they recommend it, over 570 customers gave it a perfect 5-star rating. Despite its affordable price point, a reviewer claimed that they perform just as well as more expensive brands like Snap-on, Cromwell, and Mac. Among what people liked about it, reviewers noted that they found the handles durable and comfortable.

However, there were about 3% of users that rated it a single star, including some common negative feedback about how it doesn’t hold that well. In particular, several people noted how the 90-degree pick had a tendency to bend. Dissatisfied users described how the tip had broken when used on car door trim panels, cleaning gasket surfaces, and removing oil slip seals.

Source link

Tech

Open source tool maker Grafana Labs says hackers stole its code, refuses to pay ransom

Published

29 minutes ago

18 May 2026

NewsAdmin

Grafana Labs, the maker of its eponymous popular open source web visualization software, confirmed it had been hacked but that it refused to pay the hackers who had threatened to release the company’s codebase.

In a series of posts on social media, the lab said its investigation found that the hackers had abused a stolen token credential that allowed access to the company’s GitLab environment, which it uses for code development. The token did not provide access to customer records or financial data, but allowed the hackers to obtain the company’s repositories of source code. The company has since invalidated the token and added additional security measures to prevent a repeat incident.

“The attacker attempted to blackmail us, demanding payment to prevent the release of our codebase,” the company said.

Grafana’s code is open source and public, meaning anyone can download the software and edit its code before running it on their own machines. It’s unclear if the hackers stole any proprietary code or information. A spokesperson for the company did not immediately return a request for comment.

The incident contrasts with the recent hack at education tech giant Instructure, which last week “reached an agreement” to pay the hackers who had compromised its network twice in recent weeks. The hackers had demanded an unspecified ransom, threatening to release stolen data about staff and students who use its software following a massive data breach and a subsequent website defacement.

While in Grafana’s case, no customer data was taken, the company cited the FBI’s long-standing advice urging victims not to pay hackers, as cooperating with hackers does not guarantee that they would return stolen data or refrain from publishing it later. Critics also say paying cybercriminals helps to fund future cyberattacks.

Grafana said its investigation was ongoing and will share its findings once its probe concludes.

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Source link

Tech

Agentic AI for Robot Teams

Published

42 minutes ago

18 May 2026

NewsAdmin

This presentation highlights recent efforts at the Johns Hopkins Applied Physics Laboratory to advance agentic AI for collaborative robotic teams. It begins by framing the core challenges of enabling autonomy, coordination, and adaptability across heterogeneous systems, then introduces a scalable architecture designed to support agentic behaviors in multi-robot environments. The talk concludes with key challenges encountered and practical lessons learned from ongoing research and development.

Key learnings

Source link

Tech

Running A VPN Gateway On An ESP32

Published

1 hour ago

18 May 2026

NewsAdmin

If you need a VPN gateway to access your home network, the fastest and most cost-effective way is probably by using a Raspberry Pi Zero. But in [Samir Makwana]’s view, an ESP32-S3 is just as capable for moderate use, and in some respects even superior.

This was possible thanks to the MicroLink project, which is a full implementation of a Tailscale client for the ESP32 family. In some ways the ESP32 worked better than a Raspberry Pi: it boots in two seconds rather than thirty, draws 0.5 Watts rather than 1.5, and there’s no chance of it failing due to a corrupted SD card. Compared to a Raspberry Pi, however, which can be set up as a Tailscale client in a few minutes, this took several hours to get running. The biggest issue was making sure that there was enough memory available for TLS handshakes, which was solved by enabling the ESP32’s PSRAM.

Once the VPN client is running, the ESP32 can be used as an SSH jump machine to access other devices on the home network, without needing to expose those machines to the open Internet. The ESP32 also hosts an HTTP server which can send a wake-on-LAN magic packet to another device on the local network, letting unused devices sleep without impairing their availability.

The ESP32 doesn’t provide much bandwidth — streaming video would cause issues — but it works well enough for lightweight applications. If you’re wanting to stream video from an ESP32, though, it is technically possible.

Source link

Tech

Microsoft testing adjustable taskbar, Start menu in Windows 11

Published

1 hour ago

18 May 2026

NewsAdmin

Windows 11

Microsoft has finally brought back the resizable taskbar and Start menu to Windows 11 in the latest preview version rolling out to Insiders in the Experimental channel.

Starting with Windows 11 Insider Preview Build 26300.8493, the taskbar can now be configured to use smaller buttons and moved to the bottom, top, left, or right side of the screen.

“The ability to move the taskbar to the top or sides of the screen has been one of the most requested features, and we are bringing it to Windows 11,” said Diego Baca, partner director of Microsoft Design.

“With this update, when small taskbar is enabled, you get smaller icons, a shorter taskbar, and more vertical space for your apps (see video below). No restart or sign-out is required.”

To change the taskbar position, Insiders have to go to Settings > Personalization > Taskbar > Taskbar behaviors, where they will find this new option alongside the taskbar icon alignment option.

The taskbar size can be adjusted in the same dialog by checking the “Show smaller taskbar buttons” option, which reduces the height of both the icons and the taskbar.

Microsoft is also rolling out changes to give Windows users more control over the Start menu, allowing them to toggle off recommended content and customize its size.

“These controls are designed to work together. If you want a Start menu with just your pinned apps, you can turn off Recommended and All,” Boca added. “If you want a full Start that shows everything, you can leave it all on. The goal is simple: it is your choice, and it should be easy to make.”

However, Microsoft will maintain a list of recently installed apps, as it is a key way for users to discover new applications alongside the Microsoft Store.

Furthermore, Microsoft is improving file relevance by adjusting how files are displayed and ordered to prioritize the most relevant items, and will also allow users to hide their name and profile picture from the Start menu.

Earlier this month, Microsoft also announced that it’s testing a modern Windows Run dialog with dark mode support, which should be faster than the legacy version.

This new Run dialog will also remove the “Browse” button, which was used by only 0.0038% of users in a sample of 35 million who opened Windows Run. Microsoft says the modern Run will not be enabled automatically, and users will need to turn it on manually in Settings > Advanced Settings.

Windows president Pavan Davuluri announced some of these changes in March, when Microsoft pledged to improve Windows 11 system performance and make it more responsive and consistent.

In addition to taskbar and Start menu improvements, the company plans to reduce notifications, simplify Windows settings, and ensure that device setup on new Windows PCs requires fewer reboots.

Microsoft is also working on improving Windows search, aiming for a more consistent experience across the Start menu, taskbar, File Explorer, and Settings.

“As part of this effort, we are evolving how Windows is built behind the scenes to raise the quality bar and deliver innovation where it matters most, shaped by the feedback we are hearing from you,” Davuluri said.

Automated pentesting tools deliver real value, but they were built to answer one question: can an attacker move through the network? They were not built to test whether your controls block threats, your detection rules fire, or your cloud configs hold.

This guide covers the 6 surfaces you actually need to validate.

Download Now

Source link

Tech

How to Watch Google I/O

Published

2 hours ago

18 May 2026

NewsAdmin

Google I/O is imminent. This is the annual developer event and product showcase where Google shows off all the shiny new updates to its Android operating system and other platforms, as well as new features and improvements to its artificial intelligence models.

The big announcements at Google I/O usually come in the form of a livestreamed keynote event at the Shoreline Amphitheatre in Mountain View, California, hosted by Alphabet CEO Sundar Pichai. That starts on Tuesday, May 19, at 10 am Pacific time (1 pm Eastern). You can watch the livestream on Google’s website or on YouTube. It will also be streaming right here on this page.

What to Expect

On the announcement front, Google actually tipped its hand last week, unveiling a bunch of new features coming to the upcoming Android 17 operating system and its ongoing Gemini AI efforts, which it now calls Gemini Intelligence for the most advanced Android devices.

Many of the new Android features highlighted are AI-infused tricks that aim to make your life a little easier by offloading onerous tasks like booking a hotel or ordering an Uber. Google is also bringing more robust voice-to-text recognition, some autofill smarts, and enhanced digital well-being settings that you’ll turn on for a day and then probably just disable. Also, there is a new look for Android Auto, and Google is making its emojis look more 3D. You can read more about these capabilities here.

The company also announced its new Googlebook platform: AI-first laptops powered by various Android technologies, but it’s very different from a Chromebook. Google won’t share too many specifics yet, but these machines are expected to be premium laptops, and companies like Acer, HP, Lenovo, and Dell have already signed up to make Googlebooks by year’s end. We have more details here.

We’ll likely hear Google touch on these features at I/O again, but these were unveiled early to make room for more new software announcements during the keynote. You can usually count on a surprise announcement or an AI demo that goes completely off the rails.

Based on the ever-swirling rumor mill, Google likely will announce more updates to Search, undoubtedly powered by AI. There may also be updates to its creative Google Labs suite, such as its AI music creation tool Flow. Google may introduce a new Gemini model for video generation, expected to be called Omni.

Source link

Tech

Founders Inc’s Polysynth P1 Could be the First Production Multimaterial 3D Printer

Published

2 hours ago

18 May 2026

NewsAdmin

Founders Inc. Polysynth P1 Multimaterial 3D Printer
Resin 3D printers have stuck to a single material through every layer for years because switching resins always brought contamination and extra cleanup. Eric Potempa watched that limitation long enough to do something about it. He founded Polysynth in 2025 with backing from Founders Inc and created the P1, a machine that brings up to eight different resins into the same print job without stopping for manual intervention.

The process begins with a carousel containing eight miniature circular tanks that surround the build area. Each tank contains a different type of resin, ranging from stiff plastic to super-soft flexible goo to unique conductive formulae capable of carrying an electric current. The build platform slides directly into one of the tanks, and the UV laser quickly zaps the entire layer with that neat DLP process. After that, the platform pops back up before returning to the next layer, which whips around at high speed. The centrifugal force simply flings any remaining drops of uncured resin off the part and back into the tank, a very slick trick. Then, a servo coupled to a mechanical linkage slams on the brakes, locking the platform back into place with a few microns of accuracy. It goes down into the next tank, and before you know it, you have a new layer.

Bambu Lab A1 3D Printer, Support Multi-Color 3D Printing, High Speed & Precision, Full-Auto Calibration…

High-Speed Precision: Experience unparalleled speed and precision with the Bambu Lab A1 3D Printer. With an impressive acceleration of 10,000 mm/s…
Multi-Color Printing with AMS lite: Unlock your creativity with vibrant and multi-colored 3D prints. The Bambu Lab A1 3D printers make multi-color…
Full-Auto Calibration: Say goodbye to manual calibration hassles. The A1 3D printer takes care of all the calibration processes automatically…

That cleaning spin is what allows the multi-material resin to function. Older approaches either wasted resin swirling about in large common tanks, or they required a separate washing station, which slowed everything down and increased the possibility of resin clashes. Polysynth works in the opposite way, with numerous little tanks that ensure there is always just enough resin for the job. The spin also leaves the surface immaculate and ready for the next resin, eliminating the need for any additional procedures or solvents. And because supports may often be placed directly near to the main item without contacting it, the finish is extremely smooth and easy to clean.

Dental labs are an early priority since the P1 uses biocompatible resins that meet all clinical standards. With the correct resin, you can print a full denture in one go, complete with hard tooth-colored portions pasted directly onto soft gum-like material. There is no need for any further moulding or gluing; simply drop the print in and you’re finished. The micron-level precision is perfect, just what dentists expect. The same method applies to surgical guides and personalized crowns that mix super-hard portions with flexible zones in a single component.

Founders Inc. Polysynth P1 Multimaterial 3D Printer
It’s equally as important for electronics prototyping because conductive resins can lay down entire circuits and traces within a structural housing in a single print cycle. The wiring is already incorporated in the completed item, so you can simply slap it into your project without having to solder or do any further assembly processes. Early demonstrations showed rudimentary circuit boards and flexible components manufactured simultaneously. Wearable sensors are clearly next on the list; consider rigid frames with elastic parts and built-in conducting channels, all printed in one go.

Founders Inc. Polysynth P1 Multimaterial 3D Printer
The machine starts at $4,999 and can be pre-ordered right now on the Polysynth website. Shipping dates have not been announced yet, but the business is optimistic that the P1 will be the first multi-material resin printer to reach customers. Print speeds are still fairly competitive because the DLP system zaps entire layers so quickly, and the spin cycle only adds a few seconds between resin changes. Of course, thicker resins require some tuning to get the spin just right, but the team has already worked out the most common dentistry and engineering resins.
[Source]

Source link

Tech

In the ‘AI science’ era, what is in store for the future of research?

Published

2 hours ago

18 May 2026

NewsAdmin

Sorin M.S Krammer of the University of Southampton explores the issues created by automated academic papers.

Until recently, AI’s role in research felt like having a useful assistant. It could summarise a paper, clean up a dataset or draft an abstract. Researchers were still in charge of the thinking.

That changed in late 2025 when cutting-edge “frontier” AI models became capable of reasoning and planning reliably by themselves. A key feature of these models is “tool calling” – the ability to interact with external tools in order to act on the world, not just describe it.

This marks the rise of agentic AI: systems that do not just respond to instructions but can independently plan, execute and iterate. In science as in other fields, chatbots have become coworkers that can autonomously complete real work, end to end.

An example of this is Tokyo-based Sakana AI’s The AI Scientist. Unveiled in mid-2025 and now in its second iteration, the Japanese tech company bills this as “the first comprehensive system for fully automatic scientific discovery”.

The AI Scientist scans existing literature, generates hypotheses, writes and executes code, analyses results and produces a full research paper – largely without human involvement. It reasons, fails and revises, just as a junior scientist would.

The proof? An AI Scientist academic paper was accepted in 2025 by a workshop at the International Conference on Learning Representations. This represents something genuinely new: an autonomous AI system passing a milder version of the Turing test by demonstrating scientific quality, if not (yet) machine intelligence. Moreover, the AI Scientist system was the focus of a paper published in Nature in March 2026.

Other significant achievements include Singapore-based startup Analemma carrying out a live demonstration of its Fully Automated Research System (Fars) in February. It produced 166 complete machine-learning research papers in roughly 417 hours – that’s one paper every 2½ hours – at a cost of around US$1,100 (£810) each.

Google Cloud AI Research recently unveiled PaperOrchestra, which takes a researcher’s raw experimental logs and rough notes and converts them into a submission-ready manuscript, with figures and verified citations. In blind evaluations by 11 AI researchers, it easily outperformed existing autonomous systems in this area.

Having spent two decades researching disruptive technological innovations, I believe a significant threshold has been crossed. While there is a way to go before AI systems match the very best human-produced work, the era of fully automated research has arrived.

Implications for academia

The arrival of autonomous research systems lands on an academic system under severe strain in many countries. Over the last decade, the number of papers submitted to academic journals has grown much faster than the pool of qualified peer reviewers, leading to suggestions that the science publication system is being “overwhelmed”.

If systems like Fars can produce thousands of papers per year, the publication infrastructure of science faces a volume it was never designed to handle. Some academic reviews have already been identified as using AI-generated content. As submission numbers continue to rise, this may alter the role of a published academic paper as a definitive signal of the quality and skills of human researchers.

An optimistic take is that AI may shift academia away from its strong reliance on quantity-based metrics, in favour of how influential or innovative publications are. This is a reform critics of the current system have long called for.

Less optimistically, as AI research scales up, an academic system designed for coherent, methodologically defensible contributions may inflate the proportion of incremental, rather than radically novel, scientific contributions. Both the quality and originality of research could suffer as a result.

Science has always needed its heretics to advance. Italian astronomer Galileo, the “father of modern science”, was forced to recant his defence of heliocentrism before the Catholic Church’s Inquisition. Hungarian physician Ignaz Semmelweis died in a psychiatric institution having failed to convince his colleagues that handwashing could save lives.

Yet historically, the ability of scientific institutions to encourage radical approaches has also been a mainstaple of how science has progressed. To sustain this, AI systems will need to be trained to maximise novelty and transformation, rather than plausibility and incremental progress.

AI’s impact on creative industries

The transformative effects of this new breed of AI extend well beyond scientific research. A striking example is The Epstein Files. This fully AI-generated podcast reached number one the UK Apple Podcasts and Spotify charts in early 2026, drawing 700,000 downloads in its first week.

Music is further along and more conflicted. By mid-2025, the fully AI-generated band The Velvet Sundown had amassed over a million monthly Spotify listeners. In 2026, the platform was forced to introduce artist-protection features after AI tracks began displacing human music on popular playlists, while Deezer, facing roughly 50,000 AI-generated uploads daily, began excluding them from curated lists.

Ownership remains the elephant in the room. US courts have ruled that AI-generated works cannot be copyrighted, since human authorship remains a legal requirement. AI can produce at industrial scale, but no one can own the output legally.

This matters far beyond intellectual property law. In creative industries, it threatens the royalty streams, licensing deals and catalogue valuations on which artists, labels and publishers have built their entire business models for generations.

In science, meanwhile, it is destabilising the entire incentive architecture, which rests on the foundational assumption that knowledge is both generated and owned by humans. When that assumption dissolves, so does much of the institutional logic that has governed how we produce, reward and trust expertise.

The question, across all these fields, is no longer whether AI can produce the work. Rather, it is whether sufficient thought has gone into what we will gain and lose when it does.

Sorin M.S Krammer

Sorin M.S. Krammer is a professor of strategy and international business at the University of Southampton and an Otto Mønsted visiting professor at Copenhagen Business School. His research focuses on various aspects of strategy and management in international, comparative contexts and has been previously published in outlets such as the Journal of Management, Journal of International Business Studies, Research Policy, Academy of Management Learning and Education, Journal of World Business, Journal of Product Innovation Management, Leadership Quarterly, Organisation Studies, Journal of Business Ethics and others.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

Source link

Tech

Poland builds its own Signal amid security concerns

Published

2 hours ago

18 May 2026

NewsAdmin

Security

Shift comes amid mounting reports of successful social engineering attacks targeting higher-ups in government

The Polish government is urging public officials and “entities within the National Cybersecurity System” to stop using Signal, directing them to instead use an encrypted messenger developed by a leading Polish research organization.

In an announcement on Friday, the government stated that Signal comes with security risks, including social engineering attacks orchestrated by advanced persistent threat (APT) groups.

“National-level Computer Security Incident Response Teams (CSIRTs) have identified phishing campaigns conducted by APT groups linked to hostile state agencies,” the announcement says. “These attacks target, among others, public figures and government employees.”

Offering examples of these social engineering campaigns, the government said attackers impersonate Signal support staff and abuse this perceived trust to take over victims’ accounts.

Attackers trick users into opening malicious links by sending messages designed to create a sense of urgency, such as those supposedly informing them of their account being blocked.

Successful attempts can expose victims’ phone numbers and, crucially, messages sent between government officials, potentially threatening national security.

A more detailed advisory cited “recent security incidents” related to Signal as reasons for the change.

It didn’t specify what these recent attacks were, or even who was behind them, but it can be reasonably assumed that the Polish government was indirectly referencing Russia’s phishing attempts against both Signal and WhatsApp, which were revealed in March.

Dutch intelligence agencies AIVD and MIVD reported a “large-scale” campaign targeting their own government officials, noting that some attacks were successful.

“The Russian hackers have likely gained access to sensitive information,” the AIVD and MIVD said, adding that successful attacks were carried out on government bods as well as journalists.

Beyond Signal support staff impersonation, the agencies said the attacks can also involve outsiders persuading victims to surrender their verification codes or PINs, or abusing the platform’s Linked Devices feature via QR codes to take control of accounts.

The FBI, CISA, and the German information security department issued near-identical warnings.

The alternative

Poland announced the launch of mSzyfr Messenger in March, saying it was designed for use by public administration entities, those involved in the National Cybersecurity System, and others to be decided by the government.

Developed by the Ministry of Digital Affairs and the Scientific and Academic Computer Network – National Research Institute (NASK), mSzyfr was touted by the government as “the first secure instant messenger fully under Polish jurisdiction.”

It does, however, rely on multi-factor authentication (MFA) provided by US megacorps. Microsoft is the recommended option, but users can also opt for Google or FreeOTP.

Further, if users want to retain access to messages even after logging out of the platform, they must set up a recovery key, which the installation manual suggests storing in a password manager.

That undercuts the government’s emphasis on Polish jurisdiction somewhat, since many popular password managers are either foreign-owned or open source.

An FAQ document for mSzyfr states that the messenger is built with a privacy-by-design philosophy, and explicitly notes that neither WhatsApp nor Signal fits this description. It also claimed the US-based platforms are not GDPR-compliant.

The mSzyfr app is not publicly available. Only individuals working for approved organizations are able to receive invites to join the platform.

It replaces Swiss-founded Threema, which the Polish government began endorsing for state officials and law enforcement in 2022, but data such as messages cannot be transferred because of the apps’ encrypted nature.

All Threema users should expect to receive an invite to mSzyfr in the near future, if they have not already.

The Register asked Signal to comment on Poland’s announcement, but it did not immediately respond.

It did, however, recently address security concerns raised by various intelligence agencies last week, introducing new warnings and alerts inside the platform to help users weed out potential impostors and bad actors. ®

Source link