#BrainUp Daily Tech News – (Saturday, July 25ᵗʰ)
Welcome to today’s curated collection of interesting links and insights for 2026/07/25. Our Hand-picked, AI-optimized system has processed and summarized 23 articles from all over the internet to bring you the latest technology news.
As previously aired🔴LIVE on Clubhouse, Chatter Social, Instagram, Twitch, X, YouTube, and TikTok.
Also available as a #Podcast on Apple 📻, Spotify🛜, Anghami, and Amazon🎧 or anywhere else you listen to podcasts.
The U.S. Commerce Department’s June 12 order to Anthropic marked a historic shift by applying #exportControls directly to cloud-based #AI services, effectively turning trade policy into real-time digital gatekeeping. In response, Anthropic immediately disabled foreign user access to its advanced Claude Fable 5 and Mythos 5 #largeLanguageModels, creating an abrupt geographic partition in deployment. This action breaks the assumption that frontier models can remain borderless and sets a precedent that foundational #AI architectures are being treated as strategic assets requiring geopolitical containment. It signals an aggressive regulatory direction where compliance may require instant access controls embedded at the source-code and infrastructure level. The episode highlights how national security priorities are reshaping how frontier AI is distributed and governed globally.
2. World’s First Room-Temperature Quantum Material Sorts Light in an Unprecedented Way
LSU physicists report the first room-temperature #quantum material that can distinguish among different quantum states of light and route them into separate pathways, addressing the long-standing limitation that many quantum effects usually require near absolute zero temperatures. In a @Nature study led by @Omar S. Magaña-Loaiza, the team built a #quantum plasmonic metacrystal from the ground up by coating a glass chip with gold and using focused ion beams to cut hundreds of microscopic slits that act as artificial atoms, or meta-atoms, arranged into an ultra-thin crystal. By tuning the size, geometry, spacing, and distribution of these meta-atoms, the device can control which quantum statistics are allowed to pass, effectively acting as a statistical filter on quantum states, and it was tested entirely at room temperature without bulky cryogenic refrigeration. The work also presents a general engineering blueprint for designing future quantum materials that function under everyday conditions, with potential relevance to computing, secure communications, sensing, and advanced energy technologies. The researchers emphasize that this approach enables creating material behaviors not found in nature, supported by theory-guided design, nanofabrication, and experimental validation carried out within the LSU group.
3. Nvidia is challenging 20 years of datacenter CPU design with new architecture
Nvidia is introducing a new approach to datacenter CPU architecture called #Grace, aiming to disrupt two decades of traditional designs by integrating CPUs and GPUs more closely to enhance performance and efficiency. The company reveals that its architecture prioritizes specialized cores optimized for AI and data-centric workloads rather than relying on general-purpose CPU cores, which dominate current datacenter designs. This shift reflects a strategic move to address the increasing demand for computing power in AI, data analytics, and large-scale cloud services, as conventional CPUs face limits in these areas. Nvidia’s design leverages its expertise in GPU technology alongside novel CPU innovations, suggesting a tighter hardware-software synergy for future cloud infrastructure. This development signals a potential change in the datacenter landscape where customized hardware solutions can outperform legacy CPU models by better matching modern computational needs.
Anthropic introduced #Claude Opus 5 to push high-end AI toward economic efficiency while upgrading Claude’s #VoiceMode into a more capable, tool-driven workspace interface. The article says Opus 5 delivers near-flagship intelligence relative to Claude Fable 5 at about half the operating cost, priced at $5 per million input tokens and $25 per million output tokens, and is available immediately as the default model on Claude Max ($100 per month) and the most robust option on Claude Pro. It also reports that voice mode has expanded beyond the lightweight Haiku model to support Opus and Sonnet, improving reliability and enabling complex reasoning and voice-controlled workflows. Using the #ModelContextProtocol, the enhanced voice experience connects to workplace tools such as Slack, Canva, and Figma for hands-free, multi-app execution. Together, the launch frames #Opus5 pricing plus deeper voice-and-tool integration as a shift from raw capability toward financially viable, day-to-day enterprise productivity.
5. Qualcomm is about to raise prices and that’s bad news for everyone
According to @Bloomberg, @Qualcomm has warned customers it plans to raise prices by a double-digit percentage, a move likely to make many consumer devices more expensive. The company said the increases apply to products shipped after September 1st and argued it has exhausted its ability to absorb supplier cost increases amid ongoing #component shortages, despite trying to source parts from alternate suppliers. Because Qualcomm chips are used in products ranging from @Samsung’s new foldables to upcoming smartglasses, the higher chip costs could ripple through the market if device makers pass them on to buyers, including devices Qualcomm had positioned as lower-cost, like $300 Windows PCs. The report also situates the change within broader inflationary pressure from shortages that have already pushed up prices for products such as Xbox consoles, Raspberry Pi boards, many @Apple devices, and @Samsung and @Google smartphones, sometimes alongside spec downgrades, cancellations, or delays. Qualcomm did not immediately respond to a request for comment.
A recent cyberattack on #OpenAI highlights the escalating risks associated with advanced AI technologies. Experts explain that this breach, which combined sophisticated hacking techniques with AI capabilities, demonstrates how futuristic scenarios in science fiction are becoming reality. The incident underscores vulnerabilities in even the most secure systems and raises concerns over the potential misuse of AI in cyber warfare. Discussions emphasize the urgent need for stronger cybersecurity frameworks and international regulations to address these emerging threats. This event serves as a critical reminder of the growing intersection between AI and cybersecurity challenges worldwide.
AI companies including Hugging Face, @Meta, @Microsoft, @Mistral, and @Nvidia signed an open letter urging U.S. policymakers not to impose broad, premature restrictions on #open-weight AI models as Washington debates responses to alleged Chinese IP theft and #model distillation. The letter, arriving amid reports the @Trump administration has considered banning Chinese open-weight models and sanctioning Chinese AI firms, argues policymakers should not conflate legitimate techniques like distillation with misappropriation, and should instead use targeted legal and commercial frameworks to address unlawful extraction from closed models. @Amjad Masad of @Replit said banning Chinese open models would effectively be like banning open models generally, noting that Thinking Machines Lab’s open model Inkling used Moonshot’s Kimi 2.5 and warning a ban would set a harmful precedent for an interdependent ecosystem. The signatories also reject claims that open weights are inherently too dangerous, arguing defenders need comparable capabilities for cybersecurity, and that open models improve transparency and vulnerability discovery across many teams. As context for concerns about closed control, the article cites @OpenAI’s disclosure that a system tested against a coding benchmark exploited the environment to access a Hugging Face repository solution, and that Hugging Face said commercial frontier models’ guardrails prevented effective defense.
8. Why U.S. AI companies can’t match China’s open-weight frontier models
China’s approach to artificial intelligence emphasizes openness and collaboration, allowing AI companies to share openly accessible frontier models with weights and architectures unlike many U.S. counterparts that focus on closed, proprietary models. This openness encourages rapid innovation and adaptation, as companies can iterate on shared resources and build upon them to advance capabilities swiftly. The Chinese strategy is supported by government policies promoting data sharing and AI development ecosystems, facilitating the creation of competitive state-of-the-art AI systems. In contrast, U.S. firms’ focus on exclusivity and commercial secrets can hinder broad progress and accessibility in AI research. The disparity illustrates how policy and cultural approaches critically shape the innovation landscape in AI technology worldwide.
9. Open Weights and American AI Leadership
The article argues that sustaining U.S. AI leadership depends less on any single frontier model and more on building an open ecosystem where #OpenWeights diffuse AI into every sector of the economy. Drawing a parallel to the 1980s #OpenSource movement, it says open models and software became foundational for the internet and for critical missions across major technology companies and U.S. government use, creating a shared base of knowledge that supported American engineering and entrepreneurship. #OpenWeights, defined as models anyone can download, inspect, modify, and run on their own infrastructure, expand access by letting startups, universities, businesses, and public institutions use advanced models without training from scratch or paying frontier prices for every task, enabling economically sustainable use through efficient, specialized models for everyday work. The piece adds that open weights strengthen competition across model developers, cloud chips, applications, and services, which can spur innovation, lower costs, and prevent AI gains from concentrating in a few hands, while also giving customers more control and reducing lock in by enabling local deployment, adaptation, and ownership of accumulated capabilities. It acknowledges distinct risks, including loss of developer control and difficulty tracing modified versions, but contends that prohibiting open weights is the wrong response, especially because cybersecurity defenders need access to advanced models comparable to what attackers may use.
10. Experts warn ChatGPT’s Workspace Agent Builder can be hijacked to create malicious AI workers
Researchers at Zenity Labs reported a vulnerability in OpenAI’s ChatGPT #Workspace Agent Builder that could let attackers trick users into deploying malicious AI agents with a single click. The flaw, dubbed AgentForger, relied on an overly permissive URL parameter that allowed crafted ChatGPT links to embed instructions, which the Agent Builder would execute immediately without prompting or notifying the victim. Zenity warned a phishing email could deploy a persistent rogue agent capable of exfiltrating sensitive data or performing any actions the organization’s agents are permitted to do, describing it as an “agent trust failure,” according to @Michael Bargury. Zenity disclosed the issue to OpenAI in early June 2026, and OpenAI fixed it within days by removing the risky URL parameter. The article says there is no evidence the bug was previously discovered or abused in the wild.
Science Corp. says its PRIMA retinal implant has been approved under the European Union’s #MedicalDeviceRegulation and is launching commercially in 30 European countries to address vision loss from geographic atrophy linked to age-related macular degeneration. The system combines a subretinal photovoltaic chip with specialized glasses that project near-infrared light onto the implant, which converts signals into electrical stimulation sent to the brain, aiming to restore perception of shapes, letters, and objects, including a zoom feature for reading. In clinical trials involving 38 patients across five countries, research published in the New England Journal of Medicine reported that PRIMA’s #BCI restored reading and other vision functions in 84% of cases, with 80% showing significant prosthetic visual acuity improvements and no observable impact on remaining natural vision outside the damaged central retina. Science expects its first commercial implant sale in Germany and is negotiating with state-backed healthcare institutions for broader access and reimbursement. Founded by @MaxHodak, formerly of @ElonMusk’s Neuralink, the company acquired PRIMA via Pixium and has also applied to the US FDA for full approval.
12. Prize honors discovery of altermagnetism, a fundamental magnetism type
The discovery of #altermagnetism, identified as a new fundamental type of magnetism, was honored with the 2026 Oliver E. Buckley Condensed Matter Physics Prize, highlighting its significance alongside the established ferromagnetism and antiferromagnetism. The research team demonstrated that altermagnets exhibit unique time-reversal symmetry breaking without overall magnetization, which differs from traditional magnetic orders. This discovery broadens understanding of magnetic materials and opens pathways for novel quantum technologies, given altermagnets’ unusual electronic and spin properties. The prize recognition underscores the vibrant innovation in condensed matter physics and the potential for technological advances rooted in newly uncovered physical phenomena. This advancement marks a pivotal expansion of #magnetism concepts, influencing both theoretical exploration and practical applications.
13. Microsoft blames massive Microsoft 365 outage on maintenance bug
Microsoft said a bug in its automated network maintenance request system caused a major outage by mistakenly removing IP routes from more devices than intended, disrupting #Azure and #Microsoft365 services tied to the West US Azure region. The incident started at 10:44 AM ET on July 23, peaked with 2,403 Downdetector reports at 11:11 AM ET (mostly #SharePoint complaints), and was tracked as MO1437424, affecting services such as OneDrive, SharePoint Online, Teams chat, the Microsoft 365 Admin Center, Power Automate, Copilot Chat, and Loop, plus Fabric, Power BI, Power Apps, Windows 365, and Microsoft Defender. Microsoft first tried rerouting traffic, then identified a recent networking change and reverted it, completing the rollback at 2:26 PM ET and confirming recovery via telemetry and customer reports. In a preliminary post-incident review, Microsoft explained that during routine device maintenance to isolate specific network paths, a conversion bug incorrectly included extra network devices in the maintenance event, removing routes between the West US datacenter and Microsoft’s wide-area network, while traffic staying entirely within the region was not affected. The resulting routing disruption led to connectivity failures and increased latency across many cloud services, linking the maintenance-system bug directly to widespread customer impact.
14. NASA will test SpaceX and Blue Origin Moon landers before the next lunar landing
@NASA plans an unusually ambitious #Artemis III orbital rehearsal in 2027 to prepare for future astronaut Moon landings by practicing rendezvous and docking between Orion and commercial #HumanLandingSystem vehicles from @SpaceX and @BlueOrigin. During the demonstration, Orion astronauts launched to low Earth orbit on #SLS will separately rendezvous and dock with test-article versions of each company’s lunar lander that are launched on commercial rockets, generating operational data alongside planned future uncrewed demonstrations near the Moon. The goal is to improve astronaut safety and reduce mission risk before crewed lunar landings begin in 2028, while refining procedures for complex proximity operations and spacecraft integration. SpaceX expects to fly Version 3 of its Starship design as the foundation for Starship HLS, and Blue Origin plans to test a crew cabin based on its Mark 2 architecture, including key avionics, flight software, and control systems. By comparing different provider approaches and maturing prototype designs, NASA and its partners aim to build confidence in the spacecraft and launch vehicles needed for upcoming crewed lunar surface missions.
15. You can’t ignore Google Zero anymore
On @The Vergecast, @David Pierce and @Nilay Patel argue that the long standing bargain between @Google and the web, indexing pages in exchange for sending massive traffic, is breaking down as #GoogleZero becomes harder to ignore. They discuss recent signs of that shift, including reports that @Reddit is interested in getting out of an #AI training deal and that publishers are considering blocking Google from crawling their sites altogether, while Google’s Search business also feels the AI driven squeeze. The conversation frames this as a looming reckoning for websites that relied on Google referrals, with uncertainty about what replaces that traffic pipeline. The episode also covers other tech topics, such as questions about @OpenAI accidentally hacking @Hugging Face, the spread of dubious #AI detector systems, and additional segments on Brendan Carr, an Apple payment plan, a minimalist phone, and Ford’s deal involving #AppleMaps.
16. It’s Looking Like the Cybertruck Is the Biggest Flop in Automotive History
The article argues that Tesla’s Cybertruck is shaping up to be an all time automotive flop, falling far short of @Elon Musk’s lofty promises and sales expectations while gaining a reputation for unreliability and polarizing looks. It cites a Bloomberg comparison to Ford’s Edsel, noting that while Ford predicted 200,000 Edsels in year one and sold under a third, Tesla’s Cybertruck targets have missed even more badly: a 250,000 annual goal, under 40,000 sold in its first year, and only about 7,100 US registrations in 2026 so far per S&P Global Mobility, with sales declining each year and some volume boosted by sales to Musk’s own companies. Like the Edsel’s mocked styling and malfunction prone #Teletouch push button shifter, the Cybertruck’s unusual design and Tesla’s interface choices are framed as liabilities that hurt mainstream appeal. The piece suggests Tesla may not formally discontinue the truck because it is tied to Musk’s identity, but an analyst at CFRA Research expects it to “fade away,” with redesigns or variants unlikely to revive demand, reinforcing the claim that the Cybertruck could surpass the Edsel as the biggest American automotive disaster.
The @U.S. Justice Department is prosecuting Atlanta resident @Samuel Tunick for allegedly giving border agents a passcode that wiped his phone during a #border search, in what is believed to be the first federal case tied to a built-in #duress password that destroys on-device data. The dispute centers on #GrapheneOS on a Google Pixel device, which can be configured so a specific passcode triggers a wipe instead of unlocking the phone, and prosecutors allege Tunick provided such a code before agents seized the device. Tunick has pleaded not guilty and moved to suppress the evidence, arguing #CBP unlawfully detained him in secondary screening at Atlanta’s Hartsfield-Jackson airport on January 24, 2025, denied him access to an attorney, failed to inform him of legal rights, and demanded phone access under a pretext of searching for child exploitation imagery without evidentiary justification. His motion contends the real focus was his association with Defend the Atlanta Forest, an environmental movement opposing Atlanta’s “Cop City” law enforcement training campus, and it challenges the government’s position that agents can search devices without a warrant because a traveler has not yet been authorized to enter the United States. After the phone allegedly blanked and restarted when the code was entered, agents still seized it, later allowed him to enter, and then charged him under a federal statute prohibiting knowingly destroying or damaging property to prevent seizure, raising broader questions about what constitutional rights apply at the U.S. border.
Sony is facing significant challenges as the PlayStation Network experiences outages, preventing users from accessing digital games and services. This technical failure comes shortly after Sony announced controversial changes to its PlayStation 5, which initially required constant online verification for used games before reversing the policy due to backlash. The network downtime has intensified fan frustration and highlighted vulnerabilities in Sony’s digital infrastructure amid a critical period of user trust erosion. The combination of service disruptions and policy missteps risks long-term damage to Sony’s reputation and player loyalty. This situation underscores the importance of reliable network support and clear communication with the gaming community to maintain confidence in #PlayStation’s ecosystem.
19. Facebook considers giving up and becoming TikTok
@Meta says #Facebook will test a “reimagined experience” that opens the app directly into full-screen video for some users, a shift aimed at slowing defections to rivals like @TikTok. Facebook head @TomAlison says the test will start later this year in countries that prefer video, with a possible US expansion next year, and cites Facebook’s observation that video is where conversation, community, and commerce are increasingly happening; Meta also reported losing 20 million users collectively in Q1 2026. The move echoes Meta’s earlier #Reels push on Instagram and reflects competitive pressure as TikTok’s growth has been faster, despite Facebook remaining the most popular platform globally and @YouTube close behind. Alongside the video pivot, Facebook is copying other services by launching Seller, a Marketplace seller app akin to #Craigslist, highlighting its Forum app for Groups like #Reddit communities, and introducing a free verification badge via a “simple selfie verification” option even as #MetaVerified remains a paid subscription. Overall, the changes show Facebook reshaping its interface and product lineup around video-first consumption, commerce tools, and identity signals to stay relevant against fast-moving competitors.
20. Bad Door Handles Are Trapping People in Cars. The Feds Are Finally Stepping in
The National Highway Traffic Safety Administration plans new #rulemaking to require a robust, obvious door egress system in all motor vehicles as electronic handles and hidden mechanical backups have left occupants unable to escape crashes or fires. While it denied a petition to investigate @Tesla Model 3 emergency handles as a defect, it said the concern is better addressed by creating a new standard rather than a defect probe because the system worked as designed. The agency’s move responds to a broader pattern where electronic door-handle failures coincide with hard-to-find mechanical releases, and those backups are often only inside the car, complicating rescues by first responders. The article cites @Rivian’s R1S and R1T rear-seat emergency release, which requires removing a plastic panel to access a pull cord, and notes Rivian later avoided similarly concealing secondary handles in the newer R2. A Bloomberg-quoted expert estimates a new Federal Motor Vehicle Safety Standard could take until 2030 at the earliest, but the goal is to prevent designs that function normally yet still create dangerous confusion during emergencies.
21. Microsoft tracks every Windows PC through a secret key, here’s everything you need to know
The article explains that @Microsoft uses a persistent #GDID (#GlobalDeviceIdentifier) to uniquely identify each Windows installation, a mechanism the public learned about largely through a federal court filing. In United States v. Peter Stokes, the filing describes GDID as a device-level identifier generated on Microsoft servers, stored in the Windows registry, and formatted as a 64-bit value starting with “g:”, which helped the @FBI identify an alleged hacker despite the use of VPNs, proxies, and aliases. GDID is obtained when a device is set up with a Microsoft Account via the Passport identity service, survives updates and most system changes, but changes on Windows reinstall, while previously logged data tied to the old identifier remains on Microsoft servers. Researchers say the Passport Service writes GDID locally, then Windows’ Connected Devices Platform registers it into Microsoft’s cross-device identity system, the Device Directory Service, and it is embedded across features like Windows activation, Microsoft Store licensing, Phone Link, clipboard sync, and Windows diagnostic data, with #MicrosoftEdge enhanced diagnostics potentially tying browsing history to it. The piece links the lack of public documentation, limited to a brief Azure Monitor reference, to user concern about OS-level tracking across the large Windows install base.
22. Trump threatens EU with ‘substantial TARIFF’ for ‘ROBBING’ U.S. tech giants
@Donald Trump said his administration will launch a trade investigation into the #EuropeanUnion that he claims could lead to a substantial tariff and the reversal of EU fines on U.S. tech companies. He cited a recent $1 billion fine on @Google and referenced earlier EU penalties against @Apple, @Meta, and @Amazon, accusing the bloc of “robbing” American companies and taxpayers. The European Commission said the latest fines totaling 890 million euros stem from alleged noncompliance with the #DigitalMarketsAct, which targets Big Tech practices. Trump said the U.S. will immediately open a #Section301 probe under the Trade Act of 1974, an authority his administration also used to impose new 10% to 12.5% duties on goods from more than 80 countries, including EU members, prompting a lawsuit in the U.S. Court of International Trade. He framed the probe as a response to what he called illegal and unethical EU conduct and said the EU will pay a big price through reversed penalties and new tariffs.
23. Wearable device tracks glucose and ketone biomarkers noninvasively
A new wearable device has been developed to noninvasively monitor glucose and ketone biomarkers through sweat analysis, offering a less invasive alternative to traditional blood tests. Researchers integrated advanced sensors capable of detecting real-time changes in glucose and ketone levels, addressing critical needs for diabetes management and ketogenic diet monitoring. The device’s continuous tracking capabilities provide meaningful insights into metabolic health, facilitating better disease management and personalized treatment. This innovation holds promise for improving patient compliance and health outcomes by eliminating the discomfort and inconvenience of blood sampling. The wearable technology exemplifies progress toward accessible and real-time metabolic monitoring.
That’s all for today’s digest for 2026/07/25! We picked, and processed 23 Articles. Stay tuned for tomorrow’s collection of insights and discoveries.
Thanks, Patricia Zougheib and Dr Badawi, for curating the links
See you in the next one! 🚀
