SoylentNews

Arthur T Knackerbracket has processed the following story:

An insider and an outside allegedly colluded to embezzle over $840,000 from Intel.

Israeli news source Calcalist has reported that Intel Israel has initiated legal action against Natalia Avtsin, a former employee, and Yafim Tsibolevsky, a previous component supplier, for their alleged conspiracy to embezzle over NIS 3 million, approximately $842,000. This embezzlement allegedly took place between October 2023 and November 2024, remaining undetected until Intel exposed the fraud.

Avtsin was employed in Intel Israel's hardware production department until her dismissal in November 2024. Intel stated that her termination was part of a strategy to reduce operations in Israel and was unrelated to her alleged crimes, which were still undiscovered at that time. In September 2023, Tsibolevsky registered as an authorized dealer under the name "Energy Electronics 2000" and subsequently became an official Intel supplier the following month.

Avtsin and Tsibolevsky's operation began with Avtsin asking Tsibolevsky for price quotes on hardware components. Avtsin then sent the quotes to her manager for approval, but supposedly altered the transaction classification afterward. She is said to have changed the classification from "components" to "services," which bypasses essential verification protocols. Logically, only an insider could know how a reclassification can easily bypass many security checks.

Intel Israel informed Calcalist that payments for services were less strict compared to payments for components. For service payments, a signed delivery note or confirmation receipt was not required. With no verification barriers, Tsibolevsky could submit invoices and receive payments at his convenience.

If the purchase had been classified as "services," Tsibolevsky would not have escaped scrutiny, as Energy Electronics 2000 failed to register with Intel for service provision. To evade detection, Tsibolevsky issued invoices of $20,000 or less, aligning with Avtsin's transaction limit. Once more, Tsibolevsky would have likely been unaware of this information without an insider's involvement.

Intel Israel's investigation suggests possible third-party involvement. It appears that certain transactions were processed through Levanon Kogan, a company providing purchasing services to firms that are not registered with Intel. The chipmaker has not accused Levanon Kogan of any misdemeanor, but these activities appear to correlate with the scheme of Avtsin and Tsibolevsky.

In certain operations, Avtsin obtained a quote from Tsibolevsky and requested Levanon Kogan to make payments to him. Like with other fraudulent schemes, she allegedly reclassified these transactions from components to services. Intel found that as many as 30 counterfeit orders were processed via Levanon Kogan, amounting to over NIS 2 million, or approximately $561,000.

Intel is suing Avtsin and Tsibolevsky in the Haifa District Court, asking the defendants to repay the stolen funds and any profits they made from them.

Original Submission

Arthur T Knackerbracket has processed the following story:

The Sun’s outer atmosphere—the corona—is the piping hot outer limit of our star, and is usually hidden from view except during rare total eclipses. Now, scientists have gotten their clearest look ever at this mysterious region, thanks to a new adaptive optics system that scrubs away atmospheric blur, revealing fine views of the wispy plasma on the star’s surface.

Researchers from the National Solar Observatory and New Jersey Institute of Technology unveiled the system today, along with dazzling new images and videos of the Sun’s corona. The findings, published in Nature Astronomy, show fine-scale structures in solar prominences, short-lived plasma jets called spicules, and even coronal rain: cooling plasma that falls back to the solar surface along the star’s magnetic field lines.

The team’s imaging breakthrough hinges on a technology called coronal adaptive optics. Installed on the 5.25-foot (1.6-meter) Goode Solar Telescope in California, the new system—nicknamed “Cona”—adjusts a mirror 2,200 times per second to correct for distortions caused by the churn of Earth’s atmosphere. The remarkable technology counterbalances any would-be wobble in the telescope, thereby producing particularly sharp images of the corona.

“This technological advancement is a game-changer,” said Dirk Schmidt, an adaptive optics scientist at NSO and the study’s lead author, in an observatory release. “There is a lot to discover when you boost your resolution by a factor of 10.”

Until now, solar telescopes have used adaptive optics mainly to study the Sun’s surface, the release stated. Observing the fainter corona has remained a challenge, with coronal features blurred to scales of 621 miles (1,000 kilometers)—a limit that’s existed for 80 years. But Cona now resolves features down to just 39 miles (63 km), the theoretical limit of the Goode telescope.

Among the new footage captured by the team are shots of a twisting solar prominence reshaping in real time, spicules flickering on the surface, and fine, hair-like strands of coronal rain narrower than 12.5 miles (20 km). When you consider how far the Sun is from Earth, how faint the corona is relative to the rest of the star, and how much of Earth’s turbulent atmosphere the team had to cut through and correct for, the sharpness of the images is a triumph.

“This transformative technology, which is likely to be adopted at observatories world-wide, is poised to reshape ground-based solar astronomy,” said study co-author Philip Goode, a physicist at NJIT-CSTR, in the same release. “With coronal adaptive optics now in operation, this marks the beginning of a new era in solar physics, promising many more discoveries in the years and decades to come.”

The observations offer crucial data for unraveling enduring solar mysteries—like why the corona is millions of degrees hotter than the solar surface.

The team plans to bring the coronal adaptive optics technology to the 13-foot (4-meter) Daniel K. Inouye Solar Telescope in Hawaiʻi—potentially revealing even smaller details of the Sun’s atmosphere.

Original Submission

Arthur T Knackerbracket has processed the following story:

US President Donald Trump can huff, puff, and threaten to blow Tim Cook's house down with a 25 percent iPhone import tariff, but analysts say even that threat is unlikely to bring Apple's manufacturing home.

In response to Trump's statement last week, analysts from Morgan Stanley published a research brief on Tuesday that concluded Apple is unlikely to respond to Trump's latest tariff threat in a way that will please him. 

The report, provided to The Register, concluded that the original 145 percent tariff imposed by Trump on certain imports from China last month might have made Apple budge on the matter, but since the President lost his international staredown and promised to reduce that rate, the economics no longer make sense for Cupertino.

According to the Morgan Stanley number crunchers, an iPhone manufactured in the United States would be at least 35 percent more expensive than one made overseas when accounting for tariffs on single-source components still made in China and higher US labor costs. That means a $999 iPhone would be $1,350 - at a minimum - if Apple wanted to retain a similar gross margin.

[A] 25 percent tariff will have no effect; it will need to be many times higher to compensate for the local production cost

With a 25 percent tariff on iPhone imports from China or India in place, on the other hand, Apple would need to increase prices on iPhones by only four to six percent globally to keep profits up. 

Canalys smartphone and IoT analyst Runar Bjorhovde agreed with Morgan Stanley's analysis in an email to The Register. "[A] 25 percent tariff will have no effect; it will need to be many times higher to compensate for the local production cost," Bjorhovde told us. 

Arthur T Knackerbracket has processed the following story:

The internet has seen its fair share of weird, but a Star Wars fan site secretly run by the CIA to communicate with overseas spies might top the list. StarWarsWeb.net looked like any other 2010-era fan page, complete with lightsabers, Yoda quotes ("Like these games you will"), LEGO ads, and hyped-up mentions of games like Battlefront 2 and The Force Unleashed II. But behind that nostalgic facade was a covert login system. If you entered the right password into the search bar, you'd unlock a secure line to CIA handlers. Or at least, that was the plan.

This bizarre piece of intel comes courtesy of Ciro Santilli, an independent researcher with a knack for rooting around the dusty corners of the web, who spoke to 404media.

Santili took it upon himself to dig deeper after a Reuters uncovering piece titled "America's Throwaway Spies" revealed a handful of suspicious domains back in 2022. Armed with little more than open-source tools, web dev know-how, and apparently endless patience, he ended up uncovering hundreds of similar sites.

As it turned out, the Star Wars page was just one star in a galaxy of CIA-run covert communication sites.

There were comedy pages, extreme sports sites, and even a Brazilian music fan page. Some were clearly geared toward users in states like Iran and China, where their discovery led to devastating consequences, including the execution of CIA sources around 2011-2012. But others appeared to target France, Germany, Spain, and Brazil.

The fatal flaw, according to both Santilli and the Reuters report, was that many of the sites were sloppily coded, reusing sequential IP addresses or other easily traceable breadcrumbs. Once one site was found, identifying others was often just a matter of basic detective work. This was something Iranian and Chinese counterintelligence teams apparently figured out over a decade ago. You can read this in more detail on Santilli's writeup.

Despite the deadly fallout, this digital forensics saga is now a cold case. Santilli described it as being "like a museum," saying that thanks to the Wayback Machine, people can still go back and view the site.

All said, fifteen years later, the CIA's attempt at geek-coded spycraft remains a cautionary tale that even intelligence agencies are only human. And that on the internet, your secrets have a shelf life.

Original Submission

Arthur T Knackerbracket has processed the following story:

Evidence of an attack on administration officials appeared last week on leak site Distributed Denial of Secrets, hosted an archive of messages that included details of over 60 government workers, a White House staffer, and members of the Secret.

The leak, first reported by Reuters, isn't as serious as Signalgate - no one was discussing air strikes and possible war crimes - but it's still suboptimal.

The White House said that it was "aware of the cyber security incident" but didn't comment further.

TeleMessage servers are reportedly closed while an investigation is carried out.

Europol had already detailed attempts to take down the Qakbot and Danabot malware groups, and last Friday it announced the disruption of the following five malware crews:

Operation Endgame II, a combined operation involving police from the EU, UK, US, and Canada, has now led to 20 arrests and 18 suspects have been added to the EU's most wanted list. In addition a total of €21.2 million has been seized.

"This new phase demonstrates law enforcement’s ability to adapt and strike again, even as cybercriminals retool and reorganise," said Catherine De Bolle, Europol executive director. "By disrupting the services criminals rely on to deploy ransomware, we are breaking the kill chain at its source."

Two government boffins have proposed a method for predicting which security vulnerabilities criminals are likely to exploit, and think it could be used to improve patching choices.

In a recent paper [PDF], cybersecurity specialist Jono Spring of CISA and Peter Mell, a former senior computer scientist retired from Uncle Sam's NIST this month, suggest a new system that addresses a blind spot in current flaw fixing methodologies.

Here's the current list of patches under active attack, courtesy of US government security guards at CISA.

CVSS 9.8 - CVE-2025-4632 is a path traversal vulnerability in Samsung MagicINFO 9 Server which would allow anyone with the skill to write arbitrary files as a system authority.

CVSS 7.2 - CVE-2025-4428 is a vulnerability in Ivanti Endpoint Manager Mobile 12.5.0.0 and earlier builds. It allows full remote code execution using a specially crafted API request.

One current tool to help users prioritize the fixes to deploy is the US Cybersecurity and Infrastructure Security Agency’s (CISA’s) known exploited vulnerabilities (KEV) database that lists which CVEs under active attack. Regulations require US federal government agencies to patch bugs on the list within six months. Private sector admins also use the list.

Further help comes from an industry group known as the Forum of Incident Response and Security Teams (FIRST) which feeds CVE data into a separate Exploit Prediction Scoring System (EPSS). This machine-learning system predicts which vulnerabilities criminals are likely to attack in the next 30 days.

Spring and Mell have suggested a new system to help admins that combines KEV and EPSS and called it a likely exploited vulnerabilities (LEV) list, and assert that it offers helpfully accurate indicators to focus patching priorities.

Arthur T Knackerbracket has processed the following story:

Starfish aims to roll out a battery-free system which connects to multiple parts of the brain simultaneously.

Starfish Neuroscience, a startup co-founded by Valve CEO Gabe Newell, has published an article revealing the first details of its brain-computer interface (BCI) chip.

The firm proposes a “new class of minimally-invasive, distributed neural interfaces that enable simultaneous access to multiple brain regions.” Moreover, a Starfish BCI could be the first fully wireless, battery-free implant available, if all goes to plan. According to its blog, the startup’s first chips are expected to arrive “in late 2025.” Perhaps the relationship with Newell means related tech will eventually find its way into gaming headsets and controllers.

In its report on the Starfish BCI news, The Verge notes that Newell’s fascination with BCIs began over 10 years ago, and that Valve once considered adding earlobe monitors to its VR headset products. As recently as 2019, Valve also publicly explored BCIs for gaming. Later the same year, Newell incorporated Starfish Neuroscience, and we are now seeing the first fruits as it emerges from stealth.

In its new blog post, Starfish says its BCI has the opportunity to do well thanks to two key features, its minimal size and the eschewing of built-in battery power. In regular use, the Starfish processor will consume just 1.1mW, it says. That contrasts with the Neuralink N1, which uses around 6mW.

[...] The startup also thinks that its smaller, lower power BCI implant(s) may work best connected to multiple parts of the brain simultaneously. For use in medical therapy, this multi-zone methodology could address human brain issues which affect several areas of the brain, like Parkinson’s disease.

Starfish isn’t so bold as to think it can go it alone with its new processor and BCI system. Rather, its blog floats the idea of collaborators on wireless power delivery and communication, and on custom implanted neural interfaces. It also admits “there is tons of work yet to be done here,” and is looking for employees, as well as partners, to boost its fortunes.

Original Submission

Arthur T Knackerbracket has processed the following story:

Texas could become the next US state to lay down the law with social media platforms. A Texas bill that would ban social media use for anyone under 18 recently moved past the Senate committee and is due for a vote in front of the Texas State Senate. The bill has until the state's legislative session comes to an end on June 2, leaving roughly a week for it to be approved by both the Senate and the governor.

Earlier this year, the bill passed the House committee stage and was later voted in favor of by the state's House of Representatives. If made into law, the bill would force social media platforms to verify the age of anyone setting up an account, much like how Texas passed legislation requiring websites hosting porn to implement an age verification system. On top of that, Texas' social media ban proposes to let parents delete their child's social media account, allowing the platforms 10 days to comply with the request or face a fine from the state's attorney general.

Texas isn't the only governing body interested in restricting social media access. Last year, Florida's governor, Ron DeSantis, signed into law a bill that outright bans anyone under 14 from using social media and requires 14- and 15-year-olds to get parental consent to make an account or use an existing account. Notably, Texas' proposed law is much stricter than that.

On a larger scale, the US Senate introduced a bill to ban social media platforms for anyone under 13 in April 2024. After being stuck in the committee stage, Senators Brian Schatz (D-Hawaii) and Ted Cruz (R-Texas) recently made comments that signal a potential second attempt at getting this passed.

Original Submission

Research Reveals 'Forever Chemicals' Present in Beer

hubie writes:

Research reveals 'forever chemicals' present in beer:

Infamous for their environmental persistence and potential links to health conditions, per- and polyfluoroalkyl substances (PFAS), often called forever chemicals, are being discovered in unexpected places, including beer. Researchers publishing in ACS' Environmental Science & Technology tested beers brewed in different areas around the U.S. for these substances. They found that beers produced in parts of the country with known PFAS-contaminated water sources showed the highest levels of forever chemicals.

"As an occasional beer drinker myself, I wondered whether PFAS in water supplies was making its way into our pints," says research lead Jennifer Hoponick Redmon. "I hope these findings inspire water treatment strategies and policies that help reduce the likelihood of PFAS in future pours."

PFAS are human-made chemicals produced for their water-, oil- and stain-repellent properties. They have been found in surface water, groundwater and municipal water supplies across the U.S. and the world. Although breweries typically have water filtration and treatment systems, they are not designed to remove PFAS. By modifying a U.S. Environmental Protection Agency (EPA) testing method for analyzing levels of PFAS in drinking water, Hoponick Redmon and colleagues tested 23 beers. The test subjects were produced by U.S. brewers in areas with documented water system contamination, plus popular domestic and international beers from larger companies with unknown water sources.

The researchers found a strong correlation between PFAS concentrations in municipal drinking water and levels in locally brewed beer — a phenomenon that Hoponick Redmon and colleagues say has not yet been studied in U.S. retail beer. They found PFAS in 95% of the beers they tested. These include perfluorooctanesulfonate (PFOS) and perfluorooctanoic acid (PFOA), two forever chemicals with recently established EPA limits in drinking water. Notably, the team found that beers brewed near the Cape Fear River Basin in North Carolina, an area with known PFAS pollution, had the highest levels and most diverse mix of forever chemicals, including PFOS and PFOA.

This work shows that PFAS contamination at one source can spread into other products, and the researchers call for greater awareness among brewers, consumers and regulators to limit overall PFAS exposure. These results also highlight the possible need for water treatment upgrades at brewing facilities as PFAS regulations in drinking water change or updates to municipal water system treatment are implemented.

Journal Reference: Hold My Beer: The Linkage between Municipal Water and Brewing Location on PFAS in Popular Beverages, Jennifer Hoponick Redmon, Nicole M. DeLuca, Evan Thorp, et al., Environmental Science & Technology 2025 59 (17), 8368-8379 DOI: 10.1021/acs.est.4c11265 [open access]

Arthur T Knackerbracket has processed the following story:

By combining information from many large datasets, MIT researchers have identified several new potential targets for treating or preventing Alzheimer’s disease.

The study revealed genes and cellular pathways that haven’t been linked to Alzheimer’s before, including one involved in DNA repair. Identifying new drug targets is critical because many of the Alzheimer’s drugs that have been developed to this point haven’t been as successful as hoped.

Working with researchers at Harvard Medical School, the team used data from humans and fruit flies to identify cellular pathways linked to neurodegeneration. This allowed them to identify additional pathways that may be contributing to the development of Alzheimer’s.

“All the evidence that we have indicates that there are many different pathways involved in the progression of Alzheimer’s. It is multifactorial, and that may be why it’s been so hard to develop effective drugs,” says Ernest Fraenkel, the Grover M. Hermann Professor in Health Sciences and Technology in MIT’s Department of Biological Engineering and the senior author of the study. “We will need some kind of combination of treatments that hit different parts of this disease.”

Matthew Leventhal PhD ’25 is the lead author of the paper, which appears today in Nature Communications.

Over the past few decades, many studies have suggested that Alzheimer’s disease is caused by the buildup of amyloid plaques in the brain, which triggers a cascade of events that leads to neurodegeneration.

A handful of drugs have been developed to block or break down these plaques, but these drugs usually do not have a dramatic effect on disease progression. In hopes of identifying new drug targets, many scientists are now working on uncovering other mechanisms that might contribute to the development of Alzheimer’s.

“One possibility is that maybe there’s more than one cause of Alzheimer’s, and that even in a single person, there could be multiple contributing factors,” Fraenkel says. “So, even if the amyloid hypothesis is correct — and there are some people who don’t think it is — you need to know what those other factors are. And then if you can hit all the causes of the disease, you have a better chance of blocking and maybe even reversing some losses.”

hendrikboom writes:

https://techxplore.com/news/2025-05-google-ads-ai-chatgpt.html

Google said Wednesday it is beginning to weave advertisements into its new AI Mode for online search, a strategic move to counter the challenge posed by ChatGPT as the primary source for online answers.

[...] "The future of advertising fueled by AI isn't coming—it's already here," stated Vidhya Srinivasan, Google's vice president of Ads & Commerce.

"We're reimagining the future of ads and shopping: Ads that don't interrupt, but help customers discover a product or service."

Will this make Google's so-called AI summaries better?
Will you start or continue ignoring them?
Are Google searches your preferred destination when you want to buy something?

Original Submission

The Sun is Producing Strong Solar Flares, Creating Blackouts. What to Know

upstart writes:

The sun is producing strong solar flares, creating blackouts. What to know:

A recent period of strong solar flares is expected to gradually decline over the coming weeks and months, scientists say, along with the potential for brief communication blackouts as the sun's solar cycle begins to fade.

The most powerful eruption of 2025 so far was observed last week by NASA's Solar Dynamics Observatory and the U.S. National Oceanic and Atmospheric Administration (NOAA).

The flare, classified as an X2.7, caused a 10-minute period of "degraded communications" for high-frequency radio systems in the Middle East, according to NOAA's Space Weather Prediction Center.

"We are at solar maximum, so there can be periods of more activity," a spokesperson for the Space Weather Prediction Center told Global News in an email.

The spokesperson added that the active region last week's flare emanated from, however, "has weakened magnetically, and even though it remains capable of producing a notable event, it seems less likely at this time."

[...] The 10-minute blackout in the Middle East occurred because that part of the Earth was facing the sun at the time.

However, because the active region was still somewhat off to the side, a related coronal mass ejection — which produces plasma and magnetic energy from the sun's corona — did not impact Earth.

Taylor Cameron, a space weather forecaster at the Canadian Hazards Information Service, told Global News it's difficult to predict specifically when a solar flare can erupt and which part of Earth it can affect.

The sun is currently at the peak of its 11-year solar cycle, known as solar maximum.

Although activity is generally declining, the Space Weather Prediction Center spokesperson told Global News that "sunspot activity and solar event expectations remain elevated this year and perhaps even into 2026."

Arthur T Knackerbracket has processed the following story:

The call came into the help desk at a large US retailer. An employee had been locked out of their corporate accounts.

But the caller wasn't actually a company employee. He was a Scattered Spider criminal trying to break into the retailer's systems - and he was really good, according to Jon DiMaggio, a former NSA analyst who now works as a chief security strategist at Analyst1.

Scattered Spider is a cyber gang linked to SIM swapping, fake IT calls, and ransomware crews like ALPHV. They've breached big names like MGM and Caesars, and despite arrests, keep evolving. They're tracked by Mandiant as UNC3944, also known as Octo Tempest.

DiMaggio listened in on this call, which was one of the group's recent attempts to infiltrate American retail organizations after hitting multiple UK-based shops. He won't name the company, other than to say it's a "big US retail organization." This attempt did not end with a successful ransomware infection or stolen data.

"But I got to listen to the phone calls, and those guys are good," DiMaggio told The Register. "It sounded legit, and they had information to make them sound like real employees."

Scattered Spider gave the help desk the employee's ID and email address. DiMaggio said he suspected the caller first social-engineered the employee to obtain this data, "but that is an assumption."

"The caller had all of their information: employee ID numbers, when they started working there, where they worked and resided," DiMaggio said. "They were calling from a number that was in the right demographic, they were well-spoken in English, they looked and felt real. They knew a lot about the company, so it's very difficult to flag these things. When these guys do it, they're good at what they do."

hubie writes:

Agent mode arrives, for better or worse:

Microsoft's GitHub Copilot can now act as a coding agent, capable of implementing tasks or addressing posted issues within the code hosting site.

What distinguishes a coding agent from an AI assistant is that it can iterate over its own output, possibly correcting errors, and can infer tasks that have not been specified to complete a prompted task.

But wait, further clarification is required. Having evidently inherited Microsoft's penchant for confusing names, the GitHub Copilot coding agent is not the same thing as the GitHub Copilot agent mode, which debuted in February.

Agent mode refers to synchronous (real-time) collaboration. You set a goal and the AI helps you get there. The coding agent is for asynchronous work – you delegate tasks, the coding agent then sets off on its own to do them while you do other things.

"Embedded directly into GitHub, the agent starts its work when you assign a GitHub issue to Copilot," said Thomas Dohmke, GitHub CEO, in a blog post provided to The Register ahead of the feature launch, to coincide with this year's Microsoft Build conference.

"The agent spins up a secure and fully customizable development environment powered by GitHub Actions. As the agent works, it pushes commits to a draft pull request, and you can track it every step of the way through the agent session logs."

Basically, once given a command, the agent uses GitHub Actions to boot a virtual machine. It then clones the relevant repository, sets up the development environment, scours the codebase, and pushes changes to a draft pull request. And this process can be traced in session log records.

Available to Copilot Enterprise and Copilot Pro+ users, Dohmke insists that agents do not weaken organizational security posture because existing policies still apply and agent-authored pull requests still require human approval before they're run.

By default, the agent can only push code to branches it has created. As a further backstop, the developer who asked the agent to open a pull request is not allowed to approve it. The agent's internet access is limited to predefined trusted destinations and GitHub Actions workflows require approval before they will run.

With GitHub as its jurisdiction, Copilot's agent interactions can be used to automate various development-related site interactions via github.com, in GitHub Mobile, or through the GitHub CLI.

But the agent can also be configured to work with MCP (model context protocol) servers in order to connect to external resources. And it can respond to input beyond text, thanks to vision capabilities in the underlying AI models. So it can interpret screenshots of desired design patterns, for example.

"With its autonomous coding agent, GitHub is looking to shift Copilot from an in-editor assistant to a genuine collaborator in the development process," said Kate Holterhoff, senior analyst at RedMonk, in a statement provided by GitHub. "This evolution aims to enable teams to delegate implementation tasks and thereby achieve a more efficient allocation of developer resources across the software lifecycle."

GitHub claims it has used the Copilot code agent in its own operations to handle maintenance tasks, freeing its billing team to pursue features that add value. The biz also says the Copilot agent reduced the amount of time required to get engineers up to speed with its AI models.

GitHub found various people to say nice things about the Copilot agent. We'll leave it at that.

Original Submission

slon writes:

Positive proof-of-concept experiments may lead to the world's first treatment for celiac disease:

An investigational treatment for celiac disease effectively controls the condition—at least in an animal model—in a first-of-its-kind therapeutic for a condition that affects approximately 70 million people worldwide.

Currently, there is no treatment for celiac disease, which is caused by dietary exposure to gluten, a protein in wheat, barley and rye. The grains can produce severe intestinal symptoms, leading to inflammation and bloating.

Indeed, celiac disease is the bane of bread and pasta lovers around the world, and despite fastidiously maintaining a gluten-free eating plan, the disease can still lead to social isolation and poor nutrition, gastroenterologists say. It is a serious autoimmune disorder that, when left unaddressed, can cause malnutrition, bone loss, anemia, and elevated cancer risk, primarily intestinal lymphoma.

Now, an international team of scientists led by researchers in Switzerland hope to change the fate of celiac patients for the better. A series of innovative experiments has produced "a cell soothing" technique that targets regulatory T cells, the immune system components commonly known as Tregs.

The cell-based technique borrows from a form of cancer therapy and underlies a unique discovery that may eventually lead to a new treatment strategy, data in the study suggests.

"Celiac disease is a chronic inflammatory disorder of the small intestine with a global prevalence of about 1%," writes Dr. Raphaël Porret, lead author of the research published in Science Translational Medicine.

"The condition is caused by a maladapted immune response to cereal gluten proteins, which causes tissue damage in the gut and the formation of autoantibodies to the enzyme transglutaminase," continued Porret, a researcher in the department of Immunology and Allergy at the University of Lausanne.

Working with colleagues from the University of California, San Francisco, as well as at the Norwegian Celiac Disease Research Center at the University of Oslo, Porret and colleagues have advanced a novel concept. They theorize that a form of cell therapy, based on a breakthrough form of cancer treatment, might also work against celiac disease.

coolgopher writes:

https://www.technologyreview.com/2025/05/20/1116327/ai-energy-usage-climate-footprint-big-tech/

AI's integration into our lives is the most significant shift in online life in more than a decade. Hundreds of millions of people now regularly turn to chatbots for help with homework, research, coding, or to create images and videos. But what's powering all of that?

[...] Given the direction AI is headed—more personalized, able to reason and solve complex problems on our behalf, and everywhere we look—it's likely that our AI footprint today is the smallest it will ever be. According to new projections published by Lawrence Berkeley National Laboratory in December, by 2028 more than half of the electricity going to data centers will be used for AI. At that point, AI alone could consume as much electricity annually as 22% of all US households.

[...] Racks of servers hum along for months, ingesting training data, crunching numbers, and performing computations. This is a time-consuming and expensive process—it's estimated that training OpenAI's GPT-4 took over $100 million and consumed 50 gigawatt-hours of energy, enough to power San Francisco for three days. It's only after this training, when consumers or customers "inference" the AI models to get answers or generate outputs, that model makers hope to recoup their massive costs and eventually turn a profit.

"For any company to make money out of a model—that only happens on inference," says Esha Choukse, a researcher at Microsoft Azure who has studied how to make AI inference more efficient.

As conversations with experts and AI companies made clear, inference, not training, represents an increasing majority of AI's energy demands and will continue to do so in the near future. It's now estimated that 80–90% of computing power for AI is used for inference.

Original Submission