An anonymous reader quotes a report from IEEE Spectrum: Worried that your latest ask to a cloud-based AI reveals a bit too much about you? Want to know your genetic risk of disease without revealing it to the services that compute the answer? There is a way to do computing on encrypted data without ever having it decrypted. It's called fully homomorphic encryption, or FHE. But there's a rather large catch. It can take thousands -- even tens of thousands -- of times longer to compute on today's CPUs and GPUs than simply working with the decrypted data. So universities, startups, and at least one processor giant have been working on specialized chips that could close that gap. Last month at the IEEE International Solid-State Circuits Conference (ISSCC) in San Francisco, Intel demonstrated its answer, Heracles, which sped up FHE computing tasks as much as 5,000-fold compared to a top-of the-line Intel server CPU.

Startups are racing to beat Intel and each other to commercialization. But Sanu Mathew, who leads security circuits research at Intel, believes the CPU giant has a big lead, because its chip can do more computing than any other FHE accelerator yet built. "Heracles is the first hardware that works at scale," he says. The scale is measurable both physically and in compute performance. While other FHE research chips have been in the range of 10 square millimeters or less, Heracles is about 20 times that size and is built using Intel's most advanced, 3-nanometer FinFET technology. And it's flanked inside a liquid-cooled package by two 24-gigabyte high-bandwidth memory chips—a configuration usually seen only in GPUs for training AI.

In terms of scaling compute performance, Heracles showed muscle in live demonstrations at ISSCC. At its heart the demo was a simple private query to a secure server. It simulated a request by a voter to make sure that her ballot had been registered correctly. The state, in this case, has an encrypted database of voters and their votes. To maintain her privacy, the voter would not want to have her ballot information decrypted at any point; so using FHE, she encrypts her ID and vote and sends it to the government database. There, without decrypting it, the system determines if it is a match and returns an encrypted answer, which she then decrypts on her side. On an Intel Xeon server CPU, the process took 15 milliseconds. Heracles did it in 14 microseconds. While that difference isn't something a single human would notice, verifying 100 million voter ballots adds up to more than 17 days of CPU work versus a mere 23 minutes on Heracles.

AT&T plans to invest more than $250 billion over the next five years to expand U.S. telecom infrastructure for the AI age. The company says it will also hire thousands of technicians while partnering with AST SpaceMobile to extend coverage to remote areas. Reuters reports: Rapid adoption of artificial intelligence, cloud computing and connected devices has prompted telecom operators to invest heavily in fiber and 5G networks as they also seek to fend off intensifying competition from cable broadband providers. AT&T, which has about 110,000 employees in the U.S., said the new hires will help build and maintain its infrastructure. The outlay includes capital expenditure and other spending, the company said.

The spending will focus on expanding its fiber and wireless networks, including accelerating deployment of fiber broadband, 5G home internet and satellite connectivity to extend coverage across urban, suburban and rural areas. [...] AT&T is also working with satellite partner AST SpaceMobile to expand connectivity to remote regions where traditional network infrastructure is difficult to deploy. The company said it would continue spending on the FirstNet network built for first responders and bolster investment in network security and artificial intelligence-driven threat detection.

Astronomers have spotted a galaxy they believe is made of 99.9% dark matter, reports CNN — and it's so faint, it's almost invisible: CDG-2, which is about 300 million light-years from Earth, appears to be so rich in dark matter that it could belong to a hypothesized subset of low surface brightness galaxies called "dark galaxies," which are believed to contain few or no stars.... [Post-doctoral astrophysics/statistics fellow Dayi Li at the University of Toronto was lead author on a study about the discovery, and tells CNN] There is no strict definition of dark galaxies... but their existence is predicted by dark matter theories and cosmological simulations. "Where exactly do we draw the line in terms of how many stars they should have is still ambiguous, because not everything in astronomy is as clear-cut as we like," he said. "To be technically correct, CDG-2 is an almost-dark galaxy. But the importance of CDG-2 is that it nudges us much closer to getting to that truly dark regime, while previously we did not think a galaxy this faint could exist."

To observe CDG-2, the researchers used data from three telescopes — Hubble, the European Space Agency's Euclid space observatory and the Subaru Telescope in Hawaii — along with a novel approach that involved looking for objects called globular clusters. "These are very tight, spherical groupings of very olds stars, basically the relics of the first generation of star formation," Li said. Globular clusters are bright even if the surrounding galaxy is not, and previous observations have shown a relationship between them and the presence of dark matter in a galaxy, Li added. Because CDG-2 appears to have very few stars, there must be something else providing the mass that the clusters need to hold themselves together. Li and his colleagues assume that the source of the mass is dark matter.

The researchers found a set of four globular clusters in the Perseus Cluster, a group of thousands of galaxies immersed in a cloud of gas and one of the most massive objects in the universe. Further observations revealed a glow or halo around the globular clusters, suggesting the presence of a galaxy... Astronomers believe, Li explained, that after the formation of the clusters early in the galaxy's existence, larger surrounding galaxies stripped it of the hydrogen gas required to make more individual stars like our sun. "The material that this galaxy needed to continue to form stars was no longer there, so it was left with basically just a dark matter halo and the four globular clusters." The process, he added, would leave behind a skeleton or ghost of "a galaxy that pretty much just failed." As a result of this formation mechanism, the galaxy only has 0.005% of the brightness of our own galaxy, Li said...

Studying potential dark galaxies is important because they provide nearly pristine views of the behavior of dark matter, according to Neal Dalal, a researcher at the Perimeter Institute for Theoretical Physics in Waterloo, Ontario, Canada, who was not involved with the study.
Robert Minchin, an astronomer at New Mexico's National Radio Astronomy Observatory, told CNN that "it seems likely that other very dark galaxies will be found by this method in the future."

"Seagate says it is now shipping its Mozaic 4+ HAMR-based hard drives at up to 44TB per drive," writes Slashdot reader BrianFagioli, "with production deployments already underway at two hyperscale cloud providers.

"The company claims the platform is the only heat-assisted magnetic recording [HAMR] implementation currently operating at scale, and it is targeting a path from today's 4+TB per disk toward 10TB per disk, eventually enabling 100TB-class drives." In a one-exabyte deployment, Seagate estimates Mozaic could improve infrastructure efficiency by roughly 47% compared to standard 30TB drives, cutting both footprint and energy consumption... HAMR uses a tiny laser to heat the disk surface during writes, allowing higher recording density without sacrificing stability. With most major cloud storage providers reportedly qualified on the Mozaic platform, Seagate is positioning spinning disks, not flash, as the long-term answer for cost-effective AI-scale data growth.

An anonymous reader quotes a report from 80 Level: Microsoft has officially confirmed development of its next-generation Xbox console, currently known internally as Project Helix. While concrete details remain limited, early information suggests the company is positioning the device as a hybrid between a traditional console and a gaming PC, capable of running both Xbox titles and PC games. The codename was revealed recently by new Xbox CEO Asha Sharma, who reaffirmed Microsoft's continued commitment to dedicated gaming hardware despite speculation that the company might shift entirely toward cloud or platform-based ecosystems. According to Sharma, Project Helix represents the next step in Xbox's console strategy.

Although official specifications have not yet been announced, early reports indicate the system will likely rely on a new AMD system-on-chip combining Xbox hardware with PC-style architecture. The device is expected to emphasize high performance while maintaining compatibility with existing Xbox game libraries. [...] If the concept holds, Project Helix could mark a significant shift in how console ecosystems are structured, moving away from tightly closed hardware platforms toward something closer to a unified PC-console environment. Sharma wrote in a post on X: "Great start to the morning with Team Xbox, where we talked about our commitment to the return of Xbox, including Project Helix, the code name for our next generation console. Project Helix will lead in performance and play your Xbox and PC games. Looking forward to chatting about this more with partners and studios at my first GDC next week!"

Iranian state media said on Wednesday that it targeted Amazon's data center in Bahrain due to the company's support of the U.S. military. The drone strike that occurred on Sunday disrupted core cloud services and caused "prolonged" outages. Two data centers in the UAE were also damaged by drone strikes. CNBC reports: All of the facilities remain offline, according to the Amazon Web Services health dashboard. The attack in Bahrain was launched "to identify the role of these centers in supporting the enemy's military and intelligence activities," Iran's Fars News Agency said on Telegram.

In addition to structural damage, the data centers also experienced power disruptions and some water damage after firefighters worked to put out sparks and fire. Some popular AWS applications experienced "elevated error rates and degraded availability" due to the incident. AWS advised cloud customers to back up their data, consider migrating their workloads to other regions and direct traffic away from Bahrain and the UAE.

joshuark writes: The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added a VMware Aria Operations vulnerability tracked as CVE-2026-22719 to its Known Exploited Vulnerabilities catalog, flagging the flaw as exploited in attacks. VMware Aria Operations is an enterprise monitoring platform that helps organizations track the performance and health of servers, networks, and cloud infrastructure. The flaw has now been added to the CISA's Known Exploited Vulnerabilities (KEV) catalog, with the U.S. cyber agency requiring federal civilian agencies to address the issue by March 24, 2026. Broadcom said it is aware of reports indicating the vulnerability is exploited in attacks but cannot confirm the claims.

"A malicious unauthenticated actor may exploit this issue to execute arbitrary commands which may lead to remote code execution in VMware Aria Operations while support-assisted product migration is in progress," the advisory explains. Broadcom released security patches on February 24 and also provided a temporary workaround for organizations unable to apply the patches immediately. The mitigation is a shell script named "aria-ops-rce-workaround.sh," which must be executed as root on each Aria Operations appliance node. There are currently no details on how the vulnerability is being exploited in the wild, who is behind it, and the scale of such efforts.

Intel has formally unveiled its Xeon 6+ "Clearwater Forest" data-center processor with up to 288 cores, built on the company's new Intel 18A process and using Foveros Direct packaging. The chip targets telecom, cloud, and edge-AI workloads with massive parallelism, large caches, and high-bandwidth DDR5-8000 memory. Tom's Hardware reports: Intel's Xeon 6+ processors with up to 288 cores combine 12 compute chiplets containing 24 energy-efficient Darkmont cores per tile that are produced using 18A manufacturing technology, two I/O tiles made on Intel 7 production node, as well as three active base tiles made on Intel 3 fabrication process. The compute tiles are stacked on top of the base dies using Intel's Foveros Direct 3D technology, whereas lateral connections are enabled by Intel's EMIB bridges.

Intel's 'Darkmont' efficiency cores have received rather meaningful microarchitectural upgrades. Each core integrates a 64 KB L1 instruction cache, a broader fetch and decode pipeline, and a deeper out-of-order engine capable of tracking more in-flight operations. The number of execution ports has also been increased in a bid to improve both scalar and vector throughput under heavily threaded server workloads.

From a cache hierarchy standpoint, the design groups cores into four-core blocks that share approximately 4 MB of L2 cache per block. As a result, the aggregate last-level cache across the full package surpasses 1 GB, roughly 1,152 MB in total. This unusually large pool is intended to keep data close to hundreds of active cores and reduce dependence on external memory bandwidth, which in turn is meant to both increase performance and lower power consumption. Platform-wise, the processor remains drop-in compatible with the current Xeon server socket, so the CPU has 12 memory channels that support DDR5-8000, 96 PCIe 5.0 lanes with 64 lanes supporting CXL 2.0.

At Mobile World Congress, Cristiano Amon of Qualcomm argued that the coming 6G networks will power an AI-driven "agent economy," where devices and AI assistants constantly communicate across the network. "AI will fundamentally change our mobile experiences," Qualcomm chief executive, Cristiano Amon says. "It's going to change how we think about our smartphones. Think about our personal computing. Think about and interact with a car. The car is now a computing surface. If you actually believe in the AI revolution, 6G will be required. Resistance is futile." The company says early consumer testing could begin around the 2028 Los Angeles Olympics, with broader rollouts expected by 2029. Fortune's Kamal Ahmed reports: Akash Palkhiwala is Qualcomm's chief financial officer and chief operating officer. I spent some time with him at the company's stand, as his leading engineers took me through a 6G future where individuals will have real-time information delivered to them via their glasses. Palkhiwala compliments me on my watch, which only does one thing. It tells me the time. "6G is going to be the first time that connectivity and AI come together in the network. What we're building is the first AI-native wireless network that's ever been built," he explains.

"The traffic that we expect on 6G is way different than what we had before," says Palkhiwala. "Before, it was all about consumer traffic. We expect 6G to be driven by [AI] agent traffic. Think about all these use cases where there are AI agents sitting on various devices -- your glasses, your watch, your phone, your PC. These agents are going to be talking back and forth across the network to other agents and services. "The traffic completely changes. 6G is being built with this idea that the traffic that goes on the network is not just going to be consumer voice calls or downloading videos, we're going to have agents talking to each other, so the reliability of the network becomes very important."

On-device capabilities (the ability of your phone to process far more data); edge computing (locally sourced IT technology rather than distant data centers); more efficient use of available bandwidth (AI-enabled load control); and greater cloud access will all come together to produce a new wireless network. [...] "Today we are in the application economy," he notes. "On the phone, you want to make a travel reservation, you go to one application. You want to order an Uber, you go to a second application. You want to order food, you go to a third application, movie tickets, etc. The user has to go through that effort. In the future, you think of the app economy moving over to an agent economy, where there's one agent I'm interacting with, and I can ask that agent to book me a movie ticket or a plane ticket, to order food for me, get an Uber for me. It knows everything about me."

Security researchers say a highly sophisticated iPhone exploitation toolkit dubbed "Coruna," which possibly originated from a U.S. government contractor, has spread from suspected Russian espionage operations to crypto-stealing criminal campaigns. Apple has patched the exploited vulnerabilities in newer iOS versions, but tens of thousands of devices may have already been compromised. An anonymous reader quotes an excerpt from Wired's report: Security researchers at Google on Tuesday released a report describing what they're calling "Coruna," a highly sophisticated iPhone hacking toolkit that includes five complete hacking techniques capable of bypassing all the defenses of an iPhone to silently install malware on a device when it visits a website containing the exploitation code. In total, Coruna takes advantage of 23 distinct vulnerabilities in iOS, a rare collection of hacking components that suggests it was created by a well-resourced, likely state-sponsored group of hackers.

In fact, Google traces components of Coruna to hacking techniques it spotted in use in February of last year and attributed to what it describes only as a "customer of a surveillance company." Then, five months later, Google says a more complete version of Coruna reappeared in what appears to have been an espionage campaign carried out by a suspected Russian spy group, which hid the hacking code in a common visitor-counting component of Ukrainian websites. Finally, Google spotted Coruna in use yet again in what seems to have been a purely profit-focused hacking campaign, infecting Chinese-language crypto and gambling sites to deliver malware that steals victims cryptocurrency.

Conspicuously absent from Google's report is any mention of who the original surveillance company "customer" that deployed Coruna may have been. But the mobile security company iVerify, which also analyzed a version of Coruna it obtained from one of the infected Chinese sites, suggests the code may well have started life as a hacking kit built for or purchased by the US government. Google and iVerify both note that Coruna contains multiple components previously used in a hacking operation known as "Triangulation" that was discovered targeting Russian cybersecurity firm Kaspersky in 2023, which the Russian government claimed was the work of the NSA. (The US government didn't respond to Russia's claim.)

Coruna's code also appears to have been originally written by English-speaking coders, notes iVerify's cofounder Rocky Cole. "It's highly sophisticated, took millions of dollars to develop, and it bears the hallmarks of other modules that have been publicly attributed to the US government," Cole tells WIRED. "This is the first example we've seen of very likely US government tools -- based on what the code is telling us -- spinning out of control and being used by both our adversaries and cybercriminal groups." Regardless of Coruna's origin, Google warns that a highly valuable and rare hacking toolkit appears to have traveled through a series of unlikely hands, and now exists in the wild where it could still be adopted -- or adapted -- by any hacker group seeking to target iPhone users. "How this proliferation occurred is unclear, but suggests an active market for 'second hand' zero-day exploits," Google's report reads. "Beyond these identified exploits, multiple threat actors have now acquired advanced exploitation techniques that can be re-used and modified with newly identified vulnerabilities."

sizzlinkitty shares a Reuters report detailing how drone strikes in the Middle East conflict with Iran damaged AWS data centers in the UAE and Bahrain, disrupting core cloud services and causing "prolonged" outages. Following the initial report, where Reuters said "objects" had triggered a fire at the data centers, the article was updated with additional information: A strike on the UAE facility marks the first time a major U.S. tech company's data center has been disrupted by military action. It raises questions around Big Tech's pace of expansion in the region. "In the UAE, two of our facilities were directly struck, while in Bahrain, a drone strike in close proximity to one of our facilities caused physical impact to our infrastructure," Amazon's cloud unit Amazon Web Services (AWS) said in an update on its status page. "These strikes have caused structural damage, disrupted power delivery to our infrastructure, and in some cases required fire suppression activities that resulted in additional water damage," AWS said. "We are working to restore full service availability as quickly as possible, though we expect recovery to be prolonged given the nature of the physical damage involved," it added.

Financial institutions that use AWS services have been affected by the outage, one person with direct knowledge of the situation told Reuters, requesting anonymity because of the sensitivity of the matter. "Even as we work to restore these facilities, the ongoing conflict in the region means that the broader operating environment in the Middle East remains unpredictable," AWS said. The AWS outage disrupted a dozen core cloud services and the company advised customers to back up critical data and shift operations to servers in unaffected AWS regions. Abu Dhabi Commercial Bank said its platforms and mobile app were unavailable due to a region-wide IT disruption, although it did not directly link the outage to the AWS incident. "In previous conflicts, regional adversaries such as Iran and its proxies targeted pipelines, refineries, and oil fields in Gulf partner states. In the compute era, these actors could also target data centers, energy infrastructure supporting compute, and fiber chokepoints," Washington-based think tank Center for Strategic and International Studies said last week.

Apple has reportedly asked Google to look into "seting up servers" for a Gemini-powered upgrade to Siri that meets Apple's privacy standards. The Verge reports: Apple had already announced in January that Google's Gemini AI models would help power the upgraded version of Siri it delayed last year, but The Information's report indicates Apple might lean even more on Google so it can catch up in AI.

The original partnership announcement said that "the next generation of Apple Foundation Models will be based on Google's Gemini models and cloud technology," and that the models would "help power future Apple Intelligence features," including "a more personalized Siri." While the announcement noted that Apple Intelligence would "continue to run on Apple devices and Private Cloud Compute," it didn't specify if the new Siri would run on Google's cloud. Apple's Private Cloud Compute is not only underpowered but it's also underutilized in its current state, notes 9to5Mac, "with the company only using about 10% of its capacity on average, leading to some already-manufactured Apple servers to be sitting dormant on warehouse shelves."

Saturday afternoon Sam Altman announced he'd start answering questions on X.com about OpenAI's work with America's Department of War — and all the developments over the past few days. (After that department's negotions had failed with Anthropic, they announced they'd stop using Anthropic's technology and threatened to designate it a "Supply-Chain Risk to National Security". Then they'd reached a deal for OpenAI's technology — though Altman says it includes OpenAI's own similar prohibitions against using their products for domestic mass surveillance and requiring "human responsibility" for the use of force in autonomous weapon systems.)

Altman said Saturday that enforcing that "Supply-Chain Risk" designation on Anthropic "would be very bad for our industry and our country, and obviously their company. We said [that] to the Department of War before and after. We said that part of the reason we were willing to do this quickly was in the hopes of de-esclation.... We should all care very much about the precedent... To say it very clearly: I think this is a very bad decision from the Department of War and I hope they reverse it. If we take heat for strongly criticizing it, so be it."

Altman also said that for a long time, OpenAI was planning to do "non-classified work only," but this week found the Department of War "flexible on what we needed..." Sam Altman: The reason for rushing is an attempt to de-escalate the situation. I think the current path things are on is dangerous for Anthropic, healthy competition, and the U.S. We negotiated to make sure similar terms would be offered to all other AI labs.

I know what it's like to feel backed into a corner, and I think it's worth some empathy to the Department of War. They are... a very dedicated group of people with, as I mentioned, an extremely important mission. I cannot imagine doing their work. Our industry tells them "The technology we are building is going to be the high order bit in geopolitical conflict. China is rushing ahead. You are very behind." And then we say "But we won't help you, and we think you are kind of evil." I don't think I'd react great in that situation. I do not believe unelected leaders of private companies should have as much power as our democratically elected government. But I do think we need to help them.

Question: Are you worried at all about the potential for things to go really south during a possible dispute over what's legal or not later on and be deemed a supply chain risk...?

Sam Altman: Yes, I am. If we have to take on that fight we will, but it clearly exposes us to some risk. I am still very hopeful this is going to get resolved, and part of why we wanted to act fast was to help increase the chances of that...

Question: Why the rush to sign the deal ? Obviously the optics don't look great.

Sam Altman: It was definitely rushed, and the optics don't look good. We really wanted to de-escalate things, and we thought the deal on offer was good.

If we are right and this does lead to a de-escalation between the Department of War and the industry, we will look like geniuses, and a company that took on a lot of pain to do things to help the industry. If not, we will continue to be characterized as as rushed and uncareful. I don't where it's going to land, but I have already seen promising signs. I think a good relationship between the government and the companies developing this technology is critical over the next couple of years...

Question: What was the core difference why you think the Department of War accepted OpenAI but not Anthropic?

Sam Altman: [...] We believe in a layered approach to safety — building a safety stack, deploying FDEs [embedded Forward Deployed Engineers] and having our safety and alignment researcher involved, deploying via cloud, working directly with the Department of War. Anthropic seemed more focused on specific prohibitions in the contract, rather than citing applicable laws, which we felt comfortable with. We feel that it it's very important to build safe system, and although documents are also important, I'd clearly rather rely on technical safeguards if I only had to pick one...

I think Anthropic may have wanted more operational control than we did...

Question: Were the terms that you accepted the same ones Anthropic rejected?

Sam Altman: No, we had some different ones. But our terms would now be available to them (and others) if they wanted.

Question: Will you turn off the tool if they violate the rules?

Sam Altman: Yes, we will turn it off in that very unlikely event, but we believe the U.S. government is an institution that does its best to follow law and policy. What we won't do is turn it off because we disagree with a particular (legal military) decision. We trust their authority.
Questions were also answered by OpenAI's head of National Security Partnerships (who at one point posted that they'd managed the White House response to the Snowden disclosures and helped write the post-Snowden policies constraining surveillance during the Obama years.) And they stressed that with OpenAI's deal with Department of War, "We control how we train the models and what types of requests the models refuse." Question: Are employees allowed to opt out of working on Department of War-related projects?

Answer: We won't ask employees to support Department of War-related projects if they don't want to.

Question: How much is the deal worth?

Answer: It's a few million $, completely inconsequential compared to our $20B+ in revenue, and definitely not worth the cost of a PR blowup. We're doing it because it's the right thing to do for the country, at great cost to ourselves, not because of revenue impact...

Question: Can you explicitly state which specific technical safeguard OpenAI has that allowed you to sign what Anthropic called a 'threat to democratic values'?

Answer: We think the deal we made has more guardrails than any previous agreement for classified AI deployments, including Anthropic's. Other AI labs (including Anthropic) have reduced or removed their safety guardrails and relied primarily on usage policies as their primary safeguards in national security deployments. Usage policies, on their own, are not a guarantee of anything. Any responsible deployment of AI in classified environments should involve layered safeguards including a prudent safety stack, limits on deployment architecture, and the direct involvement of AI experts in consequential AI use cases. These are the terms we negotiated in our contract.
They also detailed OpenAI's position on LinkedIn: Deployment architecture matters more than contract language. Our contract limits our deployment to cloud API. Autonomous systems require inference at the edge. By limiting our deployment to cloud API, we can ensure that our models cannot be integrated directly into weapons systems, sensors, or other operational hardware...

Instead of hoping contract language will be enough, our contract allows us to embed forward deployed engineers, commits to giving us visibility into how models are being used, and we have the ability to iterate on safety safeguards over time. If our team sees that our models aren't refusing queries they should, or there's more operational risk than we expected, our contract allows us to make modifications at our discretion. This gives us far more influence over outcomes (and insight into possible abuse) than a static contract provision ever could.

U.S. law already constrains the worst outcomes. We accepted the "all lawful uses" language proposed by the Department, but required them to define the laws that constrained them on surveillance and autonomy directly in the contract. And because laws can change, having this codified in the contract protects against changes in law or policy that we can't anticipate.

joshuark shares a report from Ars Technica: Perplexity has introduced "Computer," a new tool that allows users to assign tasks and see them carried out by a system that coordinates multiple agents running various models. The company claims that Computer, currently available to Perplexity Max subscribers, is "a system that creates and executes entire workflows" and "capable of running for hours or even months."

The idea is that the user describes a specific outcome -- something like "plan and execute a local digital marketing campaign for my restaurant" or "build me an Android app that helps me do a specific kind of research for my job." Computer then ideates subtasks and assigns them to multiple agents as needed, running the models Perplexity deems best for those tasks. The core reasoning engine currently runs Anthropic's Claude Opus 4.6, while Gemini is used for deep research, Nano Banana for image generation, Veo 3.1 for video production, Grok for lightweight tasks where speed is a consideration, and ChatGPT 5.2 for "long-context recall and wide search."

This kind of best-model-for-the-task approach differs from some competing products like Claude Cowork, which only uses Anthropic's models. All this happens in the cloud, with prebuilt integrations. "Every task runs in an isolated compute environment with access to a real filesystem, a real browser, and real tool integrations," Perplexity says. The idea is partly that this workflow was what some power users were already doing, and this aims to make that possible for a wider range of people who don't want to deal with all that setup.

People were already using multiple models and tailoring them to specific tasks based on perceived capabilities, while, for example, using MCP (Model Context Protocol) to give those models access to data and applications on their local machines. Perplexity Computer takes a different approach, but the goal is the same: have AI agents running tailor-picked models to perform tasks involving your own files, services, and applications. Then there is OpenClaw, which you could perceive as the immediate predecessor to this concept.

Mozilla has released Firefox 148 for Windows, macOS and Linux, bringing a new AI Settings section that lets users disable all of the browser's AI-powered features in one click and then selectively re-enable the ones they actually want, such as the local translation tool that works locally rather than in the cloud.

The update also patches more than 50 security vulnerabilities -- none known to be under active exploitation -- over half of which Mozilla classifies as high risk, including five sandbox escape flaws and eight use-after-free bugs in the JavaScript engine that could allow code execution.

An anonymous reader shares a report: Japan's Fair Trade Commission raided Microsoft Japan's offices on Wednesday as part of an investigation into whether it improperly restricted customers of its Azure platform from using rival cloud services, a source with direct knowledge of the matter told Reuters.

The source said Japan's antitrust authorities would also be seeking clarification from Microsoft's parent company in the United States. Microsoft Japan is suspected of setting conditions that effectively shut out other services by limiting access to popular services on other cloud platforms, the source said.

An anonymous reader shares a report: Cyberattacks reached victims faster and came from a wider range of threat groups than ever last year, CrowdStrike said in its annual global threat report released Tuesday, adding that cybercriminals and nation-states increasingly relied on predictable tactics to evade detection by exploiting trusted systems.

The average breakout time -- how long it took financially-motivated attackers to move from initial intrusion to other network systems -- dropped to 29 minutes in 2025, a 65% increase in speed from the year prior. "The fastest breakout time a year ago was 51 seconds. This year it's 27 seconds," Adam Meyers, head of counter adversary operations at CrowdStrike, told CyberScoop. Defenders are falling behind because attackers are refining their techniques, using social engineering to access high-privilege systems faster and move through victims' cloud infrastructure undetected.

Microsoft has spent more than $76 billion acquiring game studios and publishers over the past few years in an attempt to turn Xbox into a Netflix-like subscription platform, and the result is that nobody -- possibly not even Microsoft -- can clearly articulate what Xbox actually is anymore, The Verge writes.

The brand started as a powerful video game console, but Game Pass and cloud gaming pushed it toward a hazier identity: the "This is an Xbox" ad campaign tried to redefine it as any device that could play Xbox games, whether a PC, a smart TV, a phone, or a Windows handheld. Microsoft then went further and started publishing its biggest franchises on PlayStation, making it one of the largest third-party publishers on a rival's platform.

Phil Spencer, who led the division for over a decade and drove the subscription pivot, announced his retirement last week, and incoming CEO Asha Sharma has pledged "the return of Xbox" -- though her memo also talks about expanding across PC, mobile, and cloud, which sounds a lot like the status quo.

Climate scientists trying to predict how much hotter the planet will get have long grappled with a surprisingly stubborn problem -- clouds, which both reflect sunlight and trap heat, account for more than half the variation between climate predictions and are the main reason warming projections for the next 50 years range from 2 to 6 degrees Celsius.

Two research groups are now racing to close that gap using AI, though they disagree sharply on method. Tapio Schneider at Caltech built CLIMA, a model that uses machine learning to optimize cloud parameters within traditional physics equations; it will be unveiled at a conference in Japan in March. Chris Bretherton at the Allen Institute for AI took a different path -- his ACE2 neural network, released in 2024, learns from 50 years of atmospheric data and largely bypasses physics equations altogether.

"A surprisingly ravenous black hole from the dawn of the universe is breaking two big rules," reports Live Science. "It's not only exceeding the 'speed limit' of black hole growth but also generating extreme X-ray and radio wave emissions — two features that are not predicted to coexist..."

"How is this rule-breaking behavior even possible? In a paper published Jan. 21 in The Astrophysical Journal, an international team of researchers observed ID830 in multiple wavelengths to find an answer...." As they attract gas and dust, this material accumulates in a swirling accretion disk. Gravity pulls the material from the disk into the black hole, but the infalling material generates radiation pressure that pushes outward and prevents more stuff from falling in. As a result, black holes are muzzled by a self-regulating process called the Eddington limit... Its X-ray brightness suggests that ID830 is accreting mass at about 13 times the Eddington limit, due to a sudden burst of inflowing gas that may have occurred as ID830 shredded and engulfed a celestial body that wandered too close. "For a supermassive black hole (SMBH) as massive as ID830, this would require not a normal (main-sequence) star, but a more massive giant star or a huge gas cloud," study co-author Sakiko Obuchi, an observational astronomer at Waseda University in Tokyo, told Live Science via email. Such super-Eddington phases may be incredibly brief, as "this transitional phase is expected to last for roughly 300 years," Obuchi added.

ID830 also simultaneously displays radio and X-ray emissions. These two features are not expected to coexist, especially because super-Eddington accretion is thought to suppress such emissions. "This unexpected combination hints at physical mechanisms not yet fully captured by current models of extreme accretion and jet launching," the researchers said in a statement. So while ID830 is launching massive radio jets, its X-ray emissions appear to originate from a structure called a corona, produced as intense magnetic fields from the accretion disk create a thin but turbulent billion-degree cloud of turbocharged particles. These particles orbit the black hole at nearly the speed of light, in what NASA calls "one of the most extreme physical environments in the universe." Altogether, ID830's rule-breaking behaviors suggest that it is in a rare transitional phase of excessive consumption — and excretion. This incredible feeding burst has energized both its jets and its corona, making ID830 shine brightly across multiple wavelengths as it spews out excess radiation.

Additionally, based on UV-brightness analysis, quasars like ID830 may be unexpectedly common, the researchers said. Models predict that only around 10% of quasars have spectacular radio jets, but these energetic objects could be significantly more abundant in the early universe than previously suggested. Most importantly, ID830 also shows how SMBHs can regulate galaxy growth in the early universe. As a black hole gobbles matter at the super-Eddington limit, the energy from its resultant emissions can heat and disperse matter throughout the interstellar medium — the gas between stars — to suppress star formation. As a result, ancient SMBHs like ID830 may have grown massive at the expense of their host galaxies.