#BrainUp Daily Tech News – (Tuesday, August 11ᵗʰ)
Welcome to today’s curated collection of interesting links and insights for 2026/08/11. Our Hand-picked, AI-optimized system has processed and summarized 40 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.
1. AI chatbots could spread Russian propaganda in Ukraine war
AI chatbots present a growing risk of spreading Russian propaganda during the Ukraine war by automating the dissemination of biased or false information. The technology’s ability to generate misleading narratives rapidly can amplify disinformation efforts without human oversight, complicating information verification. Analysts warn that adversaries might exploit #artificialintelligence to influence public opinion and escalate conflicts covertly. This technological misuse challenges traditional media literacy and demands robust countermeasures to detect AI-driven propaganda. As AI capabilities evolve, safeguarding accurate information remains critical to maintaining democratic discourse and resisting manipulation during geopolitical conflicts.
2. Vietnam’s VinSpace signs deal with Elon Musk’s SpaceX to launch its first satellites in 2027
Vietnamese aerospace company VinSpace, part of Vingroup, signed a contract with @Elon Musk’s SpaceX to launch its first satellites on a SpaceX Transporter #rideshare mission in 2027, supporting Vietnam’s push to grow its nascent space industry. VinSpace said it will develop and operate the satellites to test its technology in orbit and enable future commercial applications, though it did not disclose contract value or satellite details. The deal aligns with VinSpace’s stated plan to become a full-service aerospace company spanning satellite design and manufacturing, launch management, satellite operations, and #space-based data services. The announcement comes as Vietnam expands national space capabilities, including opening a space science and technology center in Hanoi’s Hoa Lac High-Tech Park and targeting mid-level regional space power status by 2030. It also follows Vietnam’s decision to allow SpaceX’s #Starlink satellite internet service in the country, with Starlink already accepting orders.
3. Anthropic puts hidden watermarks on Claude text under new EU rules
Under new #EU AI Act transparency requirements, @Anthropic plans to add invisible, machine-readable watermarks to text generated by its newer #Claude models. The approach is described as embedding hidden markers in output to help identify AI-generated content without altering what humans see. This positions watermarking as a compliance-oriented tool aimed at improving traceability and disclosure for AI-produced text. The change is presented as part of Anthropic’s broader plan for meeting EU transparency rules around generative AI systems. The update ties Claude’s output format directly to #EU AI Act compliance expectations.
4. ICE to Pay LexisNexis Millions for Data to Feed to Palantir
Immigration and Customs Enforcement plans to spend millions on continued access to @LexisNexis data that can be integrated into a @Palantir platform to support deportation-related work, according to newly published procurement records. The documents say the data will support “all aspects of ICE screening and vetting, lead development, and criminal analysis,” and ICE anticipates paying $6.7 million while continuing access to LexID and Accurint, tools that link “over 82 billion public and proprietary records” from more than 10,000 sources. The procurement language specifies the database must be able to #API with ICE applications including a Palantir platform, PenLink, and ICE Data Analytics, and it lists #AI requirements such as AI-driven identity inference and bulk #facialRecognition using large-scale image databases and open-source media. The records add detail to how ICE intends to combine commercial data with Palantir infrastructure like ELITE, a mapping and targeting system previously reported by 404 Media, though it is unclear whether LexisNexis data will be ingested into ELITE specifically. The plans come as the article notes ICE arrested more than 51,000 people in July, and the agencies and companies involved did not immediately respond to requests for comment.
@Sergey Brin has increased his spending against California’s proposed #billionaire tax, donating another $20 million to Build a Better California and bringing his total to more than $100 million. The article says #Prop 40 would impose a one time 5% tax on the net worth of about 200 California billionaires, with most revenue aimed at funding state healthcare programs as California’s Medicaid program could lose up to $30 billion in federal funding due to Trump budget cuts next year. It also notes other wealthy tech figures are taking steps that could reduce exposure to the tax, citing reported moves or relocations by @Mark Zuckerberg, @Travis Kalanick, @Peter Thiel, and @Larry Page, while Governor @Gavin Newsom opposes Prop 40 over economic flight concerns and argues for a national billionaires’ tax and closing the “tax free lifestyle loan.” Californians will vote on Prop 40 in November, and the Brin backed group is supporting efforts to oppose ballot measures that could limit new taxes and effectively block Prop 40. In contrast, @Jensen Huang is quoted saying he is fine paying the tax.
6. TikTok is calling US staffers back to the office 5 days a week, starting next month
TikTok will require most US employees to return to the office five days a week starting in September, shifting many hybrid workers who currently work from home at least one day weekly. Three staffers told Business Insider the policy has been in the works for months, and TikTok previously told teams like product, marketing, and ad sales in December to prepare for a full office return in 2026, while some groups such as TikTok Shop have already been on a five day schedule for over a year. The move aligns TikTok with peers like Instagram and Amazon that also enforce #five-day #return-to-office policies, even as companies like YouTube and Patreon keep #hybrid schedules and Spotify and Dropbox emphasize flexibility. TikTok has offices in about a dozen US cities and operates in the US partly through a joint venture involving parent ByteDance and investors including @Oracle, Silver Lake, and MGX, with some functions run globally and others separated under the joint venture. TikTok did not respond to a request for comment.
#YouTube is raising the #YouTubePartnerProgram entry thresholds for ad and #YouTubePremium revenue, making it harder for new creators to qualify starting February 1, 2027. The platform will keep the 1,000-subscriber requirement, but will double the viewing thresholds to 8,000 qualified watch hours in 12 months or 20 million qualified #Shorts views in 90 days, while stating current monetized partners are not affected and lower thresholds for fan funding and #YouTubeShopping remain. YouTube will also change #Shorts revenue distribution so channels must maintain 10 million qualified Shorts views in the previous 90 days to receive Shorts ad and subscription revenue, though they stay in YPP and can still earn from long-form videos, and Shorts sharing resumes automatically if the threshold is met again. Creators and commenters like @vidIQ and @DeepHumor criticized the move as raising the bar and harming small creators, while Kick used the announcement to recruit creators to its own partner program. YouTube said creators below the new requirements will still have other earning opportunities, is expanding #PremiumLite globally where Premium exists, and claims it expects to pay partners more in 2027 than 2026 after reporting over three million YPP creators and more than $100 billion paid out from 2021 to 2025 amid recent monetization controversies.
8. Apple reveals how Siri AI features work in China – likely without privacy
Apple says its new Siri #AI features in China will be powered by @Alibaba’s #Qwen models, because Chinese law requires Apple to use local AI models instead of its usual partner @Google’s #Gemini. Outside China, the article says Siri AI runs either on Apple’s #PrivateCloudCompute servers or on Google servers dedicated to Apple, with privacy requirements intended to prevent both companies from accessing user interactions. In China, the article argues those same protections are unlikely to apply, noting the legal requirement and suggesting it is meant to enable government monitoring. According to @Reuters, Apple’s updated Chinese-language guide says users who opt in can use Qwen via Siri for more detailed responses, including analysis of photos and documents, and #WritingTools can use it to create text or images from a description. A key sign privacy may differ is that Mac users reportedly must have and be signed into their own Qwen account.
9. US hires over 2,000 video gamers as air traffic controllers
The US #DepartmentOfTransportation says it has hired more than 2,000 video gamers as air traffic controllers as part of a new recruiting push. @Sean Duffy said the initiative has reached 94% of its hiring goal and represents the most controllers employed in a single year, attributing the pace to a streamlined hiring process that is getting candidates to the Academy faster. The April recruitment campaign used footage from games like #Fortnite and highlighted that many gamers bring transferable skills such as high cognitive function, multitasking, and strategy and problem solving, which the #FAA said are relevant to air traffic control. In a video shared by Duffy, prospective candidates described air traffic control as a dream career and said gaming and flight simulators helped build their interest. The program is presented as an effort to adapt hiring to younger demographics and strengthen the aviation workforce, noting that as of April there were nearly 11,000 controllers in service and more than 4,000 trainees in the pipeline.
10. Mark Zuckerberg Posts Deranged 6,500-Word Essay About Giving Everyone AI Superintelligence
@Mark Zuckerberg published a 6,500-word essay, “The Future Is For Everyone,” pitching a vision of #AI superintelligence that the article frames as serving #Meta’s interests and blunting backlash against #AI and Meta’s practices. He promotes use cases like #Meta glasses, #AI agents, #open-weights development, and argues #data centers are not harmful to communities, while promising everyone a highly capable personal agent with strong privacy and security similar to #WhatsApp encryption and access through devices like glasses. The article argues he ignores obvious downsides, including that people may not want agents managing their lives and that widespread agents will not necessarily be used for good. As evidence, it points to recent real-world behavior: agents spamming the internet, hacking companies, and a viral incident in which an Australian man’s agent allegedly hacked a gym reservation system and canceled other people’s bookings. Overall, the piece portrays Zuckerberg’s utopian framing as detached from current harms and incentives, using grand promises to justify Meta’s products and infrastructure.
11. Nvidia partners with Wall Street giants to raise $500 billion for AI buildout
@Nvidia is partnering with six major financial institutions to launch #compute financing platforms intended to raise more than $500 billion in third-party capital for #AI infrastructure. The company signed memorandums of understanding with @Apollo, @BlackRock, @Blackstone, @Brookfield, @GoldmanSachs, and @KKR, and CEO @JensenHuang said Nvidia could backstop up to $125 billion, about 25% of potential deals. The initiative reflects surging demand for AI computing capacity as governments, companies, and startups race to build data centers, with Big Tech combined AI outlays expected to surpass $730 billion this year. Nvidia said the platforms aim to broaden access to Nvidia-based infrastructure for frontier AI developers, enterprises, governments, and cloud providers while offering longer-duration, usage-linked opportunities for large asset managers and private capital firms. Nvidia did not disclose financial terms, firm-by-firm commitments, or a timetable for deploying the planned $500 billion.
12. Attacker Takes Over Zoom AI
Zoom’s agentic chatbot #ZoomMate runs in a coding environment with unrestricted HTTPS network access and no user or admin controls to lock it down, which can let an attacker establish command-and-control and exfiltrate data via a malicious Skill or indirect #prompt injection. The described chain starts when a user requests a weekly meeting report while using a Skill they uploaded, then #Zoom AI executes the Skill’s code, which repeatedly calls an attacker server for commands, runs them inside Zoom’s server-side environment, and sends results back. The attacker can issue commands to extract meeting transcripts, messages, and data from connected services such as OneDrive, Google, and other connectors, while the user receives a normal-looking report and chat output. The connection can persist even if the user clicks stop or closes Zoom, because execution continues in the agent environment on Zoom’s servers. The article argues this appears to be intended functionality rather than a specific bug, and warns that using agentic chatbots without strict network sandboxing carries risk, especially given prior research indicating malicious Skills are being uploaded to marketplaces.
13. Deep Intelligent Pharma Advances AI for Materials Science and Advanced Material Discovery
Deep Intelligent Pharma (DIP) is expanding its AI for materials science platform to help industrial R&D teams discover and optimize advanced materials faster across areas such as batteries, semiconductors, alloys, polymers, catalysts, and agrochemicals. The platform combines #AI, scientific computing, #materials informatics, and expert knowledge to predict structures and properties, screen candidates, explain performance mechanisms, and prioritize materials for lab testing, aiming to reduce the 10 to 20 year trial-and-error cycle typical of traditional materials research. DIP says it transfers life science computational methods, including structure prediction, molecular docking, surface-energy calculation, mechanism-based design, and multiscale modeling, into materials applications, and ties models to simulations and experimental feedback in a closed research loop. For #battery materials, DIP applies high-throughput simulations to screen solid electrolytes, optimize long-cycle cathodes, and model electrode interfaces, and reports that in a collaboration it screened hundreds of millions of solid-electrolyte recipes and could make screening 100 to 1,000 times faster than conventional approaches. DIP emphasizes the technology is intended to augment scientists and laboratory validation by improving decision-making, reducing unnecessary experiments, and accelerating movement from design to commercial application.
14. UK cuts internet access to Royal Navy drone cameras after China-linked traffic was found
The UK Ministry of Defence cut internet connectivity to camera subsystems fitted on Royal Navy and Royal Marines Kraken K3 Scout uncrewed surface vessels after a routine cyber assessment found repeated outbound communications to an IP address in China. Investigators described the traffic as automated “heartbeat” messages indicating the device was online, and reporting did not establish that video, mission plans, telemetry, or other sensitive data was transmitted. The MoD said its investigation found no evidence that MoD data or systems were accessed, compromised, or sent externally, and available reporting does not show the vessel’s propulsion, navigation, or core command network was affected. The vessels are procured under #ProjectBeehive, covering 20 boats under a £12.3 million contract, and the K3 Scout’s #openArchitecture allows third-party payloads and mission systems to be integrated, which can create supply-chain exposure. Kraken said the cameras came from a third-party supplier and were represented as #NDAA-compliant, but it did not identify the vendor or which non-UK components were involved, and no independent external assessor was publicly identified beyond reviews by the MoD, the Royal Navy, and Kraken.
15. Novel bacterial strain breaks down corn stalk sugars for sustainable production
A newly identified bacterial strain demonstrates the ability to break down sugars from corn stalks efficiently, offering a sustainable alternative for biofuel production. Researchers isolated this strain because corn stalks represent an abundant agricultural waste rich in complex carbohydrates that are difficult to process. By metabolizing these sugars, the bacterium can convert biomass into valuable bioproducts, potentially reducing reliance on fossil fuels. This advancement addresses the challenge of utilizing lignocellulosic biomass, promoting a circular economy in agriculture and industry. The discovery signifies a step forward in developing renewable resources through biotechnology.
16. Intel plans $15 billion stock offering to fund AI and foundry expansion
@Intel announced a proposed $15 billion underwritten common-stock offering to fund #AI-related products and expand its #foundry and manufacturing capabilities, with proceeds earmarked for general corporate purposes like capital expenditures and working capital. The company did not disclose the offering price, share count, expected net proceeds, settlement date, or resulting dilution, and underwriters have a 30-day option to sell up to $2.25 billion in additional shares, with J.P. Morgan Securities, Goldman Sachs, Morgan Stanley, and Citigroup Global Markets as joint book-runners. The raise follows stronger demand tied to #AI, as Intel said it is increasing investment in equipment, clean-room space, and substrates, and it comes alongside reported Q2 revenue of $16.1 billion, including $6.3 billion from Data Center and AI and $5.8 billion from Intel Foundry, which posted a $2.1 billion operating loss. Intel framed the offering as supporting growth while maintaining a strong balance sheet and an investment-grade rating, against a backdrop of higher capex expectations and significant existing debt and liquidity. Overall, the planned equity financing is positioned as a flexible funding source to scale #AI products, custom silicon, advanced packaging, and contract manufacturing as Intel ramps production efforts such as Intel 18A.
17. ChatGPT’s traffic surge is good news for brands. For publishers, it’s complicated
@OpenAI changed how #ChatGPT links to websites, making links more prominent and driving a major rise in outbound clicks, a shift that benefits brands more than publishers. After May 7, referrals from ChatGPT reportedly jumped 157% overnight and stayed elevated, and a Similarweb report says the share of ChatGPT referral visits landing on brand homepages rose from about 25% to 60%, with ChatGPT already accounting for 63% of all #AI-driven referral traffic. The change coincided with OpenAI rolling out advertising in AI answers, and ChatGPT increasingly uses in-line links to company homepages rather than small citation-style links, which generally attract fewer clicks. This suggests an emerging model where AI answers are designed to produce measurable clicks that support advertising, challenging the idea of a fully #zero-click future while also reshaping who captures the traffic. The result is a web that remains click-driven in key moments, but with the most valuable link placement flowing to brands while publishers are pushed into footnotes.
18. Ofgem Proposes £712,500/MW Fee for Data Center Grid Access
UK regulator @Ofgem is consulting on a refundable #Data Centre Commitment Fee and tougher delivery tests for large data center projects seeking electricity connections in Great Britain, aiming to deter non viable applications from occupying scarce grid capacity. The proposed fee, charged when a project accepts a connection offer, ranges from £237,500 to £712,500 per MW, is held until energization, and is forfeited if the project exits the queue early, equivalent to about 2.5% to 7.5% of average data center project costs, meaning a 100MW project could need £23.75m to £71.25m posted as security. @Ofgem links the reform to a surge in contracted demand offers from 41GW to 125GW between November 2024 and June 2025, with data centers cited as roughly 73GW of the consultation queue and at least 80GW of broader demand growth. The proposed queue management milestones would require evidence of an end user, procurement of long lead electrical equipment like switchgear, and proof of financial and technical capability, with network operators able to impose penalties for missed progress. The consultation is open until September 16, 2026, as part of wider #demand connections reform to prioritize projects with a credible path to connection.
Scientists at Stanford University and the Arc Institute used an @OpenAI-created AI model called #EVO to generate new synthetic viruses, a first that could speed up medical research while raising safety concerns. EVO analyzed millions of DNA sequences from animals, plants, microbes, and viruses to learn genetic patterns, then helped researchers design hundreds of new viruses, of which 16 became viable. The 16 viruses are genetically similar to Phi X-174, which only infects bacteria, so they are not expected to threaten humans, and the approach could accelerate work like finding virus-based tools against hard-to-treat bacterial infections. At the same time, a Science commentary warned that while the ability to compose viral genomes with #generativeAI now exists, the #biosafety and #biosecurity governance to steer it safely does not. The researchers argue that stronger safety measures and guardrails are needed as AI-enabled biology advances.
20. Signed up for Klaviyo? Dozens of advertisers may have seen your password | TechCrunch
Security research says marketing tech company Klaviyo inadvertently shared new-customer sign-up data, including passwords, with third-party advertisers due to a misconfigured web form. Researcher Sam Jadali of Melurna said the sign-up page was misconfigured from at least February 2024 through November 2025, potentially sending entered details such as email address, password, company name, website, and phone number to embedded tracker partners including Facebook, Google, HubSpot, Microsoft and LinkedIn, and X. Klaviyo confirmed it fixed what it called an #application configuration issue, and a spokesperson said fewer than 200 known individuals were affected based on available active logs, without clarifying log retention or how long the bug may have existed. The incident highlights how #pixel trackers can be misconfigured to capture and transmit sensitive data, and follows similar tracker-related lapses that have prompted breach disclosures and regulatory enforcement. TechCrunch reports Klaviyo says it notified known affected individuals, but the company did not provide the notice and it is unclear why the incident was not publicly disclosed.
21. Amazon Used a Startlingly Dirty Trick to Build a Data Center Without Consent of Community
Amazon began building a large #data center in Gilroy, California by relying on decades-old #zoning rules that allowed approval without meaningful public awareness or input. The Wall Street Journal reports that a single city staffer approved a 438,500-square-foot facility on 56 acres of former farmland, and residents say many only learned of it after construction started, with some discovering it years later. Locals objected because data centers can consume major resources and have been linked to higher electricity prices, increased greenhouse gas emissions, loud noise, and water impacts, while not typically producing many jobs, and one resident said she was blocked from commenting on an environmental review because the public comment window closed before she knew the project existed. Amazon’s @AWS said it followed standard processes for an industrially zoned parcel but conceded it should have sought more engagement sooner, and the company has since met residents, hosted an open house, and promoted benefits like a wastewater pipeline, tax revenue, construction jobs, and festival sponsorship. The episode has prompted city officials to consider changing approvals to require community meetings, reflecting growing conflict between rural communities and tech companies over #AI-related data center expansion.
22. One of China’s workhorse rockets just exploded in flight
China’s Long March 7A, a frequently used medium lift rocket, exploded about 1 minute and 20 seconds after a seemingly nominal liftoff from the Wenchang Space Launch Site, with the classified payload believed to be a military communications satellite. Public videos indicate the failure occurred during first stage flight, possibly near maximum dynamic pressure, and it may have involved an onboard #flight termination system if the vehicle veered off course. Chinese officials had not commented, and given the closed nature of the state run program, the precise cause may remain unclear despite China’s recent record of 50 or more launches per year with only a handful of failures. If the problem traces to the first stage @YF-100 engine, the implications could be broad because the kerosene and liquid oxygen engine powers the Long March 7 and 7A and is used across as many as nine Chinese rockets, including variants informing newer vehicles. An engine related issue could also delay the upcoming Chang’e 7 lunar mission, planned to launch on a Long March 5 whose boosters use YF-100s, aiming to send an orbiter, lander, rover, and mini hopper to the South Pole region near Shackleton crater amid a broader China NASA race toward that potentially resource rich area.
23. OpenAI announces premium business pricing as it seeks to increase revenue ahead of IPO
OpenAI is adding a premium pricing tier to #ChatGPT Business to expand revenue as it prepares for a potential #IPO. The new premium seats target power users with five times the usage of standard Business seats and cost $125 per user per month, or $100 per user per month billed annually, compared with $25 per month or $20 annually for the standard plan, matching #Anthropic’s Teams pricing. The move fits OpenAI’s broader push into business and enterprise markets as competition intensifies from @Anthropic and @Google, with Anthropic gaining large corporate customers through its Claude Code platform and both firms rolling out workplace productivity offerings like #ChatGPT Work and Claude Cowork. With both companies having filed confidential S-1s with the SEC in June and reports suggesting OpenAI could delay its offering until 2027, premium business pricing is positioned as a way to drive higher sales beyond consumer products ahead of going public.
24. Brex assumes its AI agents could do anything , so it watches the network, not the code
@Brex CEO Pedro Franceschi argues enterprises should secure AI agents by treating them like potentially untrusted “virtual employees” and enforcing control at the network boundary rather than by restricting or auditing the agent’s constantly changing code. After Brex aimed the open-source OpenClaw at internal roles, the security team rejected traditional deployment because agents have code execution capabilities and could be hard to control, prompting Brex to build #CrabTrap, an open-source HTTP proxy that monitors all outbound traffic from the agent’s container. CrabTrap assumes the agent “can do anything” and may already be compromised, so it checks whether each outbound request aligns with an approved policy, using an #LLM as a judge for risky actions. To avoid the heavy latency of judging every request with an LLM, Brex routes routine low-risk traffic through static pre-approved rules while escalating high-risk actions, such as sending emails, to the LLM, reducing the LLM path to roughly 2% of traffic. The approach reframes agent security from controlling tools or code, such as limiting capabilities in competing models like Nvidia’s NemoClaw, to governing what the agent can communicate over the network while preserving the coding flexibility that makes agents valuable.
Cambridge Aerospace has raised a $300 million Series C to scale low-cost air defence interceptors aimed at improving the cost-exchange ratio against drones and missiles. The round values the UK-based manufacturer at $3.4 billion and was led by DFJ Growth with participation from Lux Capital, Accel, Lakestar, Never Lift, Ora Global, and Elad Gil & Co, bringing total funding since its September 2024 incorporation to more than $630 million, and marking a rapid valuation jump from $1.3 billion in April 2026. Its Skyhammer interceptor targets slower threats like attack drones and cruise missiles at roughly Mach 0.7 with about 30 km range, with unit costs in the “tens of thousands” of dollars, positioned as about 1 to 2% of conventional interceptor prices by using #autonomy for guidance and emphasizing mass production. A second system, Starhammer, is a rocket-powered option intended for faster and higher-value threats up to ballistic missiles and is planned for market in 2027, while the company is also building a solid rocket motor facility in Norfolk to ease manufacturing bottlenecks. The funding will primarily expand output from low hundreds per month toward several thousand Skyhammers monthly, supported by a UK Ministry of Defence contract announced in April, successful tests in Jordan, inclusion in the LEAP programme, and ongoing talks with the US government, under a team led by @Steven Barrett.
26. Orbital-debris removal is moving from demonstrations to government-backed contracts
The commercial #orbital-debris removal market is shifting from demonstrations toward government-backed contracts, but most capabilities proven so far are inspection, rendezvous, or servicing rather than full removal of unprepared objects. Evidence includes a June low Earth orbit rocket-body fragmentation reported by the UK National Space Operations Centre, a discarded @SpaceX Falcon 9 upper stage impacting the Moon on August 5 after a long high-elliptical trajectory, and ESA modeling that warns breakups can keep growing the debris population even if launches stop. On the customer side, Starfish Space holds the clearest U.S. commercial disposal award, a Space Development Agency #SBIR Phase 3 deorbit-as-a-service contract with up to $52.5 million to dispose of up to seven satellites starting with a mission targeted for 2027, while @Astroscale completed close-range inspection of an unprepared H-IIA upper stage and has a roughly ¥13.2 billion @JAXA follow-on contract intended for capture and deorbit. The article argues that real demand is emerging but the opportunity is narrower than the total debris count because firms focus on large, trackable targets or satellites designed for servicing, and because ownership and liability rules generally prevent removing another state’s registered object without authorization. Overall, cleanup progress is increasingly tied to government programs and operator agreements rather than a fully open commercial market for legacy debris.
27. Xbox PC and Game Pass games are being worked on for Linux
Microsoft is actively working on bringing Xbox PC and Game Pass games to Linux, expanding accessibility beyond Windows users. This initiative involves adapting games to run natively or through compatibility layers on Linux systems, enhancing the gaming ecosystem for open-source platforms. The move highlights Microsoft’s commitment to broader platform support and its strategy to grow the Game Pass subscriber base by reaching more users. By enabling Game Pass on Linux, Microsoft leverages the flexibility of Linux and the growing interest in gaming outside traditional operating systems. This expansion aligns with industry trends where Linux gaming receives increased attention and development resources.
28. America is betting on small modular reactors to revive nuclear shipping
The US government is backing an international push to reduce regulatory barriers for civilian nuclear reactors on ships and floating platforms, aiming to enable offshore power plants and nuclear-powered maritime operations. The @International Atomic Energy Agency will launch its #ATLAS initiative in Washington on August 26 to 27, preceded by a US Industry Day on August 25, with @Rafael Mariano Grossi arguing that #small modular reactors can provide high energy density and long operating cycles without frequent refueling. US officials including @Chris Wright link the effort to the Department of Energy’s broader goal of expanding atomic power and “energy dominance,” while US nuclear and maritime companies anticipate a potential global market. The article notes both precedents and tradeoffs: the US Navy already uses nuclear propulsion, the civilian NS Savannah (1959) was costly to operate, and Russia’s floating plant Akademik Lomonosov has produced 70 MW since 2019. It also highlights that putting reactors at sea changes rather than removes risks, including offshore emergency access challenges, reliance on backup power, and possible marine contamination affecting nearby coastal communities.
29. Your agent didn’t hallucinate; it exceeded its authority
Enterprises are misdiagnosing many #AI agent incidents as hallucinations when the real issue is missing #decision rights, an agent can act correctly yet exceed what the business authorized. Examples include refund workflows issuing credits beyond approved autonomous limits, order agents applying changes while ignoring financing or fulfillment constraints, and procurement agents selecting low cost suppliers without defined permission to accept contractual terms. Content filters and other guardrails address unsafe or invalid output, but they cannot determine whether an agent is authorized to take an external, business binding action, creating a governance gap as teams move from copilots to tool using agents. A 2026 @Cloud Security Alliance survey reported widespread agent related incidents and unknown agents in environments, and a @World Economic Forum playbook proposes an auditable authorization profile, reinforcing the need to separate access from authority. The article argues every production agent should have a machine enforceable #Agent Authority Contract defining ownership, allowed actions, system reach, materiality limits, escalation triggers, reversibility, and expiry, so safe actions are also legitimate actions.
30. The U.S. has more than 500 data-center moratorium records. That is not 500 AI bans.
A claim that the U.S. now has more than 500 “AI data center bans” misreads @Michael Bommarito’s Moratorium Nation inventory, which counts local #moratorium instruments, not permanent AI-only prohibitions. As of July 31, the dataset logged 533 instruments across 42 states, including 435 active or extended, 35 pending or proposed, and 63 expired, replaced, or rescinded measures, spanning five categories: data centers, cryptocurrency mining, battery energy storage, utility-scale solar, and wind. In this legal sense, a moratorium is a temporary pause on accepting or approving permits while officials study impacts or draft permanent rules, often lasting 6 to 12 months, and the database tracks outcomes like replacement by regulation or rescission, while also counting each local instrument separately. Examples show varied terms, including Seattle’s emergency pause on large new data centers and Prince George’s County’s two-year hyperscale-focused pause, and the dataset notes that document coverage is uneven and some earlier counts are lower bounds due to less systematic review and limited online records. The better takeaway is that there is widespread local permitting activity affecting #data centers and related infrastructure, but the 533 figure is not evidence of 500 permanent #AI bans.
31. Samsung Attacks the AI Memory Wall with 3D Packaging – BigDATAwire
The article says the “AI memory wall” has become the biggest bottleneck for running AI inference workloads, and that @Samsung presented efforts to address it using #3D packaging. It notes that this was discussed during the Future of Memory and Storage Summit, where Samsung “threw the kitchen sink” at the problem. The framing implies that packaging and memory architecture improvements are being positioned as key levers to relieve inference performance constraints caused by memory limitations. Beyond these points, the provided text does not include specific technical details, products, benchmarks, or outcomes. In context, the piece links Samsung’s 3D packaging push directly to alleviating the AI inference memory bottleneck highlighted at the event.
32. Archer to acquire Boeing’s Wisk, Insitu and SkyGrid units in stock deal
@Archer Aviation plans to acquire @Boeing subsidiaries Wisk Aero, Insitu, and SkyGrid in a stock transaction, with @Boeing receiving a 16.5% stake in Archer, according to @CNBC. The combination would place Wisk’s autonomous electric-aircraft program and SkyGrid’s airspace-management work alongside Archer’s piloted Midnight air taxi effort, while Insitu brings unmanned aircraft and related systems. Reporting did not disclose purchase price, valuation, share terms, employee or asset transfers, closing conditions, or required regulatory approvals, and it characterizes the deal as an ongoing collaboration without identifying separate commitments for financing, manufacturing support, or government contracting. Wisk’s autonomous Gen 6 aircraft is flight-testing and pursuing #FAA type certification, including flights of its first and second Gen 6 vehicles in late 2025 and May 2026, and Wisk and SkyGrid have also published work on #AutomatedFlightRules for highly automated operations. Archer says Midnight’s type and production certification remain in progress, and it is preparing for initial U.S. Midnight operations in the second half of 2026 via the federal #eVTOLIntegrationPilotProgram, with the report not indicating that Wisk’s Gen 6 will replace or merge with Midnight.
33. Google’s classic Search button is gone in a new AI-first homepage
Google is testing an #AI-first Google Search homepage for users who are signed out, replacing the usual Google Search button with shortcuts for Create images, Ask about files, and Brainstorm, and adding an #AIMode button inside the search field. The layout was observed in the stable version of Chrome without a Google Account, and also appeared in Microsoft Edge and Perplexity’s Comet browser. In testing, Ask about files and Brainstorm worked while signed out, including a + button that lets users attach a file and ask questions about it, and #AIMode was usable without signing in. Create images differed by requiring a Google Account sign-in. The AI shortcuts disappeared after signing in and returned after signing out, and Google has not confirmed whether this is a limited test or a broader server-side rollout by account, session, or location.
34. Tech leaders say AI means less work – their staff say they work up to 90 hours a week
Tech executives promoting #AI as a path to fewer working hours are not matching those claims inside their own companies, where staff say workloads are intensifying. A @Google engineering director predicted #AI would enable a four-day work week by 2025, and @OpenAI later urged companies to test a four-day week with no pay cut, but a former OpenAI technical employee told the BBC the company never trialled it and instead operated with weekend work, frequent crisis meetings, and cut-throat performance reviews. Workers interviewed said so-called “sprints” in AI teams can run for weeks and reach 90 hours in a seven-day period at OpenAI and Anthropic, while the former OpenAI employee reported routinely working at least 70 hours a week. Leaders have also framed #AI as letting fewer employees do more, with @MarkZuckerberg saying @Meta is entering a period where AI dramatically changes work, alongside layoffs, and Anthropic’s @DarioAmodei warning higher AI productivity could mean broader job losses. The accounts suggest #AI development and deployment are currently associated with longer hours and urgency-driven reorganizations, including Meta workers describing being “drafted” onto AI teams without real choice, contradicting the promised near-term liberation from work.
35. California wants to ban AI therapists as thousands turn to chatbots for healthcare
California lawmakers are advancing #SenateBill903 to restrict #AI chatbots from acting or being marketed as therapists as use of chatbot-based mental health support rises. The bill would ban companies from advertising chatbots as therapy, require licensed professional review before AI systems make therapeutic decisions, and require patient permission for providers to use AI to record therapy sessions or triage care, with sponsor @StevePadilla arguing #AI algorithms cannot replicate human therapists’ training and may harm patients without guardrails. Opponents, including TechNet’s Robert Boykin, say the measure could stifle legitimate #AI uses and create a clinician bottleneck in intake and screening amid behavioral health worker shortages. The bill is also framed by concerns about oversight and safety, including a complaint by the National Union of Healthcare Workers alleging Kaiser Permanente used an automated e-visit system without clinician oversight, which Kaiser disputes, and by broader warnings that chatbot care needs more study and regulation. Studies cited show substantial adoption and limited standards, including a 2025 RAND, Brown, and Harvard finding that about one in eight adolescents and young adults use chatbots for mental health care, an APA study reporting 77 percent of psychologists have patients who used AI for support, and @OpenAI reporting 1.2 million weekly messages to #ChatGPT with explicit indicators of potential suicidal planning or intent, amid difficulties accessing in-person care.
36. The AI Slop Backlash Is Actually Having an Impact
The backlash against AI-generated content, often dismissed as “AI slop,” is influencing how AI tools are developed and deployed. Users and creators increasingly criticize AI outputs for lacking originality and quality, prompting companies to refine their algorithms and prioritize human oversight. This response reflects a growing awareness that AI must improve to meet creative and ethical standards. The pushback highlights the tension between rapid AI adoption and the necessity for responsible, high-quality innovation. Consequently, this dynamic shapes the future trajectory of AI, emphasizing the need for collaboration between human creativity and artificial intelligence.
Windows 11’s built-in Weather app is criticized as a clear example of the operating system’s remaining problems, namely excessive #bloat and intrusive advertising. Windows Latest reportedly saw the app use over 1.2GB of RAM, and TechRadar observed peaks around 1GB, far higher than macOS’s native Weather app at roughly 250MB, while the Windows app also shows side-panel ads for third-party products like shoes or medical insurance. The article attributes the heavy memory use to the app effectively being the MSN Weather web app, running as a web app with eight separate #Chromium processes, with ads pulled from the MSN network. This resource demand is framed as especially poorly timed amid a broader “RAM crisis” and a push to make Windows 11 more viable on 8GB PCs, and the continued presence of ads in basic default apps has drawn social media backlash, including complaints about paying for a premium OS yet still being served ads.
38. Farmer trusted AI after it gave good advice, which helped save his crops
A farmer decided to trust an AI system after it provided accurate advice that helped save his crops from disease. The AI’s recommendations, based on data analysis and environmental factors, allowed timely interventions to protect the farm’s yield. This instance highlights how AI technology can become a valuable tool in agriculture by offering precise and actionable insights. The success story shows the potential for AI to improve farming efficiency and sustainability. The growing trust in AI applications may lead to wider adoption in sectors like agriculture, benefiting productivity and resource management.
39. The ‘Manosphere’ Isn’t a Movement. It’s a Multibillion-Dollar Grievance Industry
A new report argues the so-called #manosphere is less an organic political movement than a profit-driven #grift economy that turns young men’s loneliness and anxiety into resentment and sales. Researchers at @Reset Tech and @Equimundo describe a funnel in which men searching for dating, fitness, or finance advice are algorithmically routed toward paid communities, unregulated supplements, crypto schemes, gambling, and expensive coaching pitched as “belonging” and “wealth.” The report says bad actors reframe real stressors, like labor-market competition, into grievance narratives with targets such as women, immigrants, and foreign workers, then monetize the resulting sense of victimhood through memberships, courses, and gray-market products. As an example, it estimates @Andrew Tate earns about $184 million in annual gross revenue from his $99-per-month “Real World” platform, and it characterizes such systems as “fraud at scale” marked by recycled content, affiliate pyramids, cancellation barriers, and deceptive testimonials. The overall claim is that online “masculinity” is being constructed and financially incentivized, making the ecosystem resemble a multibillion-dollar pyramid scheme more than a coherent movement.
40. Linus Torvalds says AI has made ‘huge’ Linux kernel updates the new normal
@Linus Torvalds says the #Linux kernel project has entered a “new normal” of unusually large release-candidate updates, which he attributes in part to fixes prompted by reviews from various #AI tools. In Linux 7.2 rc6 he noted it was the biggest rc6 in years by commit count, and in the rc7 announcement he said he is not thrilled by the size but that nothing looks particularly scary, just a lot of changes. He highlighted a few larger items, including s390/zcrypt fixes, the return of the btrfs fixup worker infrastructure, and netfilter ipset fixes, while most other updates are many small fixes spread across drivers, filesystems, networking, and architecture code. The increased volume reflects AI-influenced review and contribution dynamics that Torvalds has previously criticized for generating poor code and overwhelming bug reports, even as he maintains Linux is not anti-AI. Despite the bloat, he expects Linux 7.2 to release next weekend unless something really bad appears, and notes 7.2 is not an LTS release but will include features like improved GPU and CPU scheduling.
That’s all for today’s digest for 2026/08/11! We picked, and processed 40 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! 🚀
