#BrainUp Daily Tech News – (Thursday, June 11ᵗʰ)
Welcome to today’s curated collection of interesting links and insights for 2026/06/11. Our Hand-picked, AI-optimized system has processed and summarized 38 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. UK Gov’t to deploy new supercomputer by 2030, with British chips (hopefully)
The UK government’s new #AI Hardware Plan targets launching a national AI supercomputer by 2030 while backing UK chip and semiconductor development and building related skills. It earmarks £750 million for a supercomputer to be deployed at the University of Edinburgh in 2030, replacing #ARCHER2, alongside £400 million for “next-generation chips,” including £150 million to buy next-generation inference chips and £250 million for additional specialized chips, with an aim for British-designed chips to be a crucial part of the system and a tender process starting soon. The plan also funds an #AI Hardware Innovation Programme with £120 million for UK companies to design, develop, and test novel chips, including at least £20 million to expand the Scaling Inference Lab run by #ARIA and CommonAI, where Oriole Networks will deploy what the government calls the first large-scale AI system using purely photonic networking with AMD Instinct GPUs and AMD EPYC CPUs. To address workforce needs, £45 million supports doctoral training and undergraduate bursaries, including a £12 million Centre for Doctoral Training in Chip Design and expansion of the semiconductor skills program to 400 undergraduate bursaries next year. A new investment fund led by Playground Global, backed by up to £150 million from the British Business Bank and involving former Intel CEO @Pat Gelsinger, will invest in UK AI hardware companies, with Playground opening its first non-US office in the UK.
2. Anthropic CEO Warns AI Is Getting Too Powerful, While Releasing Powerful AI
In an essay titled “Policy on the AI Exponential,” @Dario Amodei says governments must move from studying AI regulation to imposing binding safety requirements on #frontier AI models because transparency alone is no longer sufficient. He argues AI progress is outpacing policy, noting rapid capability gains, and he calls for an FAA-like system in which models undergo technical testing and auditing, with releases blocked or reversed if they fail safety standards. He proposes mandatory third-party testing, government authority to stop unsafe deployments, securing model weights, required safety testing, and serious-incident reporting, along with preparations for #job displacement, limits on surveillance and autonomous weapons in domestic law enforcement, and stronger cooperation among democratic nations on critical AI technologies. The policy push comes as #Anthropic expands access to Claude Mythos with Mythos 5 for cybersecurity and government partners, which researchers including the UK’s AI Security Institute found can autonomously execute complex cyber attacks, and as it launches Claude Fable 5 with safeguards that route sensitive requests to a less capable model, a move criticized for higher token usage, 30-day data retention, and capability-reducing safeguards without user notice. Amodei frames the situation as a race between fast-moving technology and slower public policy as Anthropic prepares for an #IPO after filing with the U.S. SEC.
3. Visa and Mastercard partner with OpenAI’s ChatGPT to transform online shopping experience
Visa and Mastercard are collaborating with OpenAI’s ChatGPT to create a more intuitive and personalized online shopping experience. This partnership integrates #AI-powered conversational technology into payment services, allowing users to interact with shopping platforms through natural language. The collaboration aims to simplify purchasing decisions and enhance customer engagement by leveraging ChatGPT’s advanced language understanding capabilities. This innovation represents a broader trend in financial services to adopt cutting-edge #AI tools for improving user experience. By combining Visa and Mastercard’s payment infrastructure with OpenAI’s technology, the companies anticipate setting new standards in digital commerce.
4. The theory taking the rich by storm: China funds data center haters
A growing theory among some U.S. officials and Silicon Valley investors claims that local opposition to new #data centers is being driven by foreign, specifically Chinese, money, despite little direct evidence tying specific activists to China. Interior Secretary @Doug Burgum alleged some opposition is “foreign-sourced dark money,” and on the All-In podcast investor Gavin Baker suggested a “CCP-funded campaign” may be behind it. Concrete signs of Chinese activity described in the article center on @OpenAI, which said it banned likely Chinese accounts that used #ChatGPT to generate anti-data-center social media content about energy demand and electricity costs, but @Ben Nimmo said the effort appeared limited and did not create the underlying debate. The broader context is widespread domestic resistance, including a Gallup poll reporting 71% of Americans somewhat or strongly oppose data centers in their communities, alongside growing Washington interest such as Rep. Brett Guthrie requesting a briefing on possible Chinese influence. Overall, the article portrays a real but seemingly small influence attempt intersecting with an already-existing, largely homegrown backlash, while prominent backers amplify claims of Chinese funding without presenting direct links.
5. OpenAI says China launched influence campaign to shape US attitudes on AI data centers
@OpenAI reported that China was likely behind an online influence operation aimed at shaping U.S. perceptions of #AI and reframing Washington’s debate over #AI data centers and federal tech policy. OpenAI said it detected the operation because China-linked actors used #ChatGPT to generate social media posts, images, and comments that were then published by batches of accounts posing as Americans, and it identified two clusters of users likely originating from China. The narratives pushed included claims that data center build-outs raise electricity costs for average American families and that @DonaldTrump’s tariffs were weaponized to keep the U.S. ahead, with another cluster’s prompts written in Simplified Chinese and asking to avoid mentioning @XiJinping while focusing only on Trump; the accounts involved were banned. OpenAI’s Ben Nimmo said the campaigns drew little authentic engagement and showed intent more than impact, while amplifying existing U.S. backlash tied to environmental, energy-grid, and water-supply concerns and proposed moratoriums. The findings are likely to intensify Republican and pro-AI claims of foreign meddling in the data center debate, even as OpenAI said it has not seen evidence of widespread influence on public opinion.
6. Bloomberg – Are you a robot?
Bloomberg.com is blocking access due to detected unusual activity from the user’s computer network and requests a verification step to confirm the user is not a robot. The page instructs the user to click a checkbox and ensure the browser allows JavaScript and cookies and is not blocking them, and it references Bloomberg’s Terms of Service and Cookie Policy for more information. It also offers support contact instructions and provides a block reference ID (b03952bc-6694-11f1-a849-1bf7696ebc4d). The message ends with a prompt to subscribe to Bloomberg for global markets news.
7. OpenAI Considers Drastic Price Cuts, Anticipating War for Users with Anthropic
OpenAI is planning significant price reductions for its AI services as it anticipates increased competition with Anthropic, a rival AI startup. This move reflects the escalating battle to capture users in the growing #GenerativeAI market, where companies are striving to offer accessible and affordable AI tools. By lowering prices, OpenAI aims to retain and expand its user base amidst a surge of competitors developing advanced language models. The strategy indicates the intensifying market dynamics as leading AI firms seek dominance in shaping the future of AI applications. OpenAI’s pricing adjustments highlight the critical role cost plays in user adoption and market competition within the AI industry.
8. Inside India newsletter: Hollywood is debating AI. India’s filmmakers are embracing it
Indian filmmakers are rapidly adopting #generativeAI to produce streaming and movie content as demand rises and budgets tighten in India’s $32 billion media and entertainment sector. JioStar, a joint venture of Reliance Industries and @Walt Disney, launched a 100-episode AI-made retelling of the “Mahabharat” that drew 6.5 million views on launch day and performed 2.1 times above the platform average, and the company says it is a first step toward expanding storytelling. Other examples include Abundantia Entertainment’s planned theatrical release “Chiranjeevi Hanuman – The Eternal,” described locally as India’s first AI-generated feature film, and “Made in India: the Titan story” on Amazon’s MX Player, which used over 100 AI shots to recreate 1970s Mumbai. Creators report a surge in AI production work, with one AI artist saying he now receives 10 to 15 job offers weekly after previously having to chase work. Executives and researchers say the key advantage is faster production timelines and lower costs, making time to market critical in a content-hungry industry, supported by growing use of tools like #Seedance, #Minimax, #GoogleStudio, #Midjourney, and #AdobeFirefly.
9. Chinese agents caught rebuilding botnets and stirring the pot on AI datacenter debate
Reports say China-linked operators are both reviving infrastructure used for intrusion and testing influence campaigns tied to US #AI and #datacenter debates. Lumen Black Lotus Labs reports a “significant resurgence” of a botnet linked to China-nexus actors including #VoltTyphoon, noting that while the KV cluster was largely defunct after the FBI said it dismantled the KV-botnet in January 2024, the JDY cluster remains active and has grown to more than 1,500 compromised routers and #IoT devices, with reconnaissance quickly following public vulnerability disclosures and a prominent focus on US military associated entities. Separately, @OpenAI says it banned #ChatGPT accounts likely originating from China that used its models to generate social posts and images for covert operations about American AI, including content framing AI and datacenters as drivers of electricity demand and higher household costs, often linking to real news stories. OpenAI’s @BenNimmo said neither influence campaign appears to have gained much authentic engagement, but they are notable for revealing the narratives operators are testing. The article frames the botnet resurgence as the more pressing threat and notes the recommendation that enterprises implement #CISA and #NCSC guidance to mitigate Volt Typhoon activity and defend against China-nexus covert networks of compromised devices.
10. Microsoft doesn’t know what to do about the memory pricing crisis Microsoft is causing
Because a #memory pricing crisis that its own #AI push is helping drive is making high-end components harder to price for mass audiences, @Microsoft is struggling to position its next Xbox, codenamed Helix, as “leading-end” while keeping it affordable. New Xbox boss @AshaSharma told Fortune the industry needs “new business models” rather than simply the most premium console, and suggested the company will need to rethink #storage and #memory via compression, flexible storage offerings, and encouraging games that can “fit on-device.” The article argues these ideas are vague, that Xbox has already offered flexible storage before, and speculates the new model could involve financing, rent-to-own, or #cloudStreaming, which could reduce reliance on local RAM and SSD capacity. Sharma also wrote in an internal letter that Microsoft cannot currently make as many consoles as demand requires and needs new business models and partnerships for hardware while remaining committed to Helix, underscoring that the company has no clear plan yet. The piece links this uncertainty to broader worries about rising hardware prices, citing a $950 1TB Steam Deck and urging readers to “safeguard that #DDR5.”
11. Anthropic’s new model refuses to find smart contract vulnerabilities
The public release of @Anthropic’s Claude Fable 5, a scaled-back version of its earlier Mythos model, is drawing criticism because its #guardrails frequently block or redirect sensitive requests, including smart contract security work. Anthropic uses “classifiers” that detect topics like #cybersecurity, #biology, #chemistry, and #distillation, then route users to Claude Opus 4.8, which has led users attempting smart contract audits to report refusals and policy-violation blocks, according to Colossus Pay CEO Joseph Delong, Yearn developer Banteg, and wallet recovery founder Zeng Jiajun. Observers argue the safeguards are overly broad, with crypto security expert Taylor Monahan saying it changes little for typical security practitioners, while @HaydenAdams warns defensive hardening requests can look indistinguishable from black-hat activity. The model’s distillation protections also reportedly “nerf” requests related to frontier LLM development, and biology-related classifiers have blocked even basic interaction and questions like those about mitochondria, prompting biologist Olivia H. Scharfman to call for better classifiers. Anthropic says it prioritized releasing Fable quickly even if safeguards are overly broad, citing concerns about abuse such as bioweapons and viruses, which explains why many requests fall back to Opus 4.8.
12. Canada announces bill banning social media for anyone under 16 – Engadget
Canada has introduced the #SafeSocialMediaAct to ban anyone under 16 from having a social media account and to impose new safety obligations on social media services and certain #AI chatbot services. Introduced by @Marc Miller, the bill would require platforms to design products to be safer for children, remove deepfakes and content that sexually victimizes a child or revictimizes a survivor, and add measures such as AI content labels, clear reporting tools, and user blocking to reduce exposure to harmful material. While chatbots are not age gated, the bill adds expectations that AI platforms mitigate risks of chatbots communicating harmful content or engaging in harmful behavior, and provide emergency measures for crisis situations, with Miller referencing concerns following @OpenAI’s handling of the Tumbler Ridge shooting. Miller said the specific requirements beyond the under 16 restriction will be set by the newly formed #DigitalSafetyCommissionofCanada, which will enforce the rules and can grant exemptions if a platform has sufficient safeguards for children.
Anthropic says it is joining calls for the AI industry to find ways to cushion people from AI-driven disruptions to the workforce. The article reports that the company is pledging $200 million to research #AI’s economic impact, and notes that CEO @Dario Amodei has suggested solutions to potential job losses. This funding effort is framed as a response to concerns that #AI could significantly reshape employment and the broader economy. By backing research and discussing mitigation ideas, Anthropic positions the industry as having a role in preparing for and managing workforce disruption. The pledge ties the company’s research investment directly to debates about how to reduce the social and economic harms of #AI-related job displacement.
14. Meta’s Ray-Bans Aren’t the Only Smart Glasses With a ‘Glasshole’ Problem
Smart glasses like #camera-enabled smart glasses are creating recurring privacy problems, and @Meta’s Ray-Bans are not the only example. A report cited from the Xiaoxiang Morning Post says Rokid AI glasses, sold in the U.S., have allegedly been used in China to film people without consent, with such footage reportedly shared publicly on Rokid community forums, including videos trending on Weibo that show interactions with Spring Airlines flight attendants who were allegedly recorded unknowingly. The same report also found third party sellers offering stickers that obscure Rokid’s recording LED without triggering the glasses’ sensors that are supposed to disable recording when the light is blocked, mirroring similar attempts to hide recording indicators on Ray-Ban Meta glasses. Separate reporting by journalist @Joanna Stern is mentioned as showing people in the U.S. will drill out the light for money, making recording even more discreet. The article argues that once these devices are in the real world, it is nearly impossible to police misuse, and it remains unclear whether company damage control can persuade lawmakers and the public that such glasses should be socially and legally acceptable.
15. Palantir’s Karp says businesses are ‘unhappy’ with the frontier AI labs
@Alex Karp says Palantir’s enterprise customers are privately unhappy with how frontier AI labs operate, arguing the labs do not understand their businesses and focus on “tokenmaxxing,” burning through #AI tokens to signal productivity. He linked the dissatisfaction to rising #LLM usage in enterprise workloads, with accelerating model costs raising alarm on Wall Street and sharpening efficiency concerns. Karp said #large language models are crucial, but that most value for the next several years will come from implementation rather than the models themselves, and he claimed many of @Anthropic’s publicly discussed projects run on Palantir. His remarks come as @OpenAI and Anthropic take steps toward going public, and he also criticized the politicization of #AI, calling it a massive U.S. revolution that should not be reduced to a blue-red debate because it will drive major political decisions and carries both opportunities and dangers.
16. Apple dashed my Apple Intelligence dreams
The author argues that Apple’s marketing of the iPhone 16 Pro as “Built for Apple Intelligence” has proven misleading because major upcoming #AppleIntelligence capabilities are being reserved for newer, higher-spec devices. At WWDC 2026, @Craig Federighi said the “most powerful on-device model” and features it enables, including expressive Siri voices and more advanced dictation, will come only to Apple’s “most capable” systems, with a slide indicating requirements like iPads with M4 or later and 12GB RAM, Macs with M3 or later and 12GB RAM, and for phones an iPhone Air or iPhone 17 Pro. As a result, the author says their iPhone 16 Pro will miss key features they expected after upgrading, despite Apple’s earlier promises of “AI for the rest of us” and a significantly improved Siri. They note Apple has explicitly named only two iPhone 17 Pro-only features so far, but the phrasing suggests more limitations could follow, potentially affecting tools like Spatial Reframing and other image-generation features with heavy model demands. The author concludes that Apple’s shifting compatibility boundaries and unclear communication make the situation “indefensible” for recent flagship buyers who upgraded expecting full access to #AppleIntelligence.
A German court ruled that #Google’s #AIOverviews are treated as Google’s own content, making the company directly liable for false statements rather than benefiting from the limited liability typically granted to search engine operators. The Regional Court of Munich issued a temporary injunction (case no. 26 O 869/26) after Google’s AI overviews falsely linked two Munich-based publishers to scams, subscription traps, and dubious business practices, including claims that did not appear in any of the linked sources and appeared to result from mixing them up with other companies. The court emphasized that AI overviews do not function like traditional search results because they rewrite and evaluate information “in its own words and according to its own structure,” which undermines Google’s defense that users can verify the information themselves. Google responded that the overviews are designed to reflect existing web information and that it invests heavily in quality, while acknowledging occasional context misses or misinterpretations, but the ruling draws a clear liability line precisely because the system generates new content rather than merely listing sources. The decision signals that AI-generated summaries in search can be judged as publisher speech with corresponding responsibility for harm caused by inaccurate outputs.
18. Canada’s teen social media ban comes with a Big Tech off-ramp
Canada’s Liberal government introduced the #SafeSocialMediaAct to bar children under 16 from creating social media accounts, while offering an exemption for platforms that can demonstrate effective child-safety safeguards. @MarcMiller said the bill would impose new safety requirements on services like Instagram, X, and Snapchat, backed by steep penalties, and would require removal within 24 hours of sexually explicit content that victimizes children and adults, including AI-generated #deepfakes. Companies could qualify for a carve-out by proving guardrails against bullying and harmful content encouraging self-harm or body dysmorphia, while #AI chatbots are excluded from the ban but would face added rules, such as directing users to help when prompts involve suicide ideation, self-harm, or criminal activity. The legislation still must pass Parliament, and it is timed to bolster Canada’s push for coordinated international action on online harms and child safety at the #G7, even as France has struggled to win U.S. and U.K. support and the Trump administration has warned against foreign regulation of U.S. tech firms. An industry group, the #ChamberOfProgress, urged the government to drop the ban.
19. macOS 27 beta boots Asahi Linux off Apple Silicon
The macOS 27 beta prevents #AsahiLinux from booting on #AppleSilicon by making its Linux boot volume stop appearing in Apple’s boot picker and the Startup Disk app. The Asahi team says the beta changed how those tools detect valid OS boot volumes, so the Asahi partition is no longer visible even though it still exists and data is not lost. They advise users not to upgrade to macOS 27 until a fix lands, and if someone insists on trying it, to keep a secondary install of macOS 26 or put macOS 27 on a secondary volume; the installer has also been updated to block installs on macOS 27 for now. The team has filed a bug report and notes the issue is likely accidental given the OS is in beta, while warning they will not support users who installed the macOS 27 beta without ensuring a stable macOS fallback. Overall, it is presented as a temporary setback for Linux on Apple hardware rather than the end of the project, which recently released Fedora Asahi Remix 44.
20. Ukrainian AI Interceptors Are Now Downing Russian Shahed Drones Autonomously
Ukraine is fielding #AI-enabled interceptor drones that can shoot down Russian Shahed attack UAVs with 95% of the interception process automated from launch to impact. Developed by MaXon Systems within the Brave1 defense tech cluster, the system plugs into Ukraine’s radar network for real-time tracking, then an operator selects a target and authorizes a strike while the software autonomously guides the interceptor and its onboard AI identifies and locks onto the Shahed. A #human-in-the-loop safeguard remains, letting the operator cancel at any moment. Brave1 says the technology has passed combat trials in the Kharkiv region, and that investment support helped move it from prototype to battlefield deployment in about a year. The rollout fits a broader strategy to build more autonomous air defense as Russia increases Shahed launches by about 35% each month, and @Mykhailo Fedorov said Shahed shootdowns by specialized interceptor drones have doubled over the past four months.
21. SpaceX pitches Starlink as telecom infrastructure
In its IPO filing, SpaceX frames #Starlink less as consumer satellite broadband and more as a broad telecom connectivity platform spanning enterprise networking, government communications, and carrier partnerships. It reports its Connectivity segment generated $11.4 billion in 2025 revenue, nearly triple the $4.1 billion from its launch business, driven by subscriber gains, rising enterprise adoption, and improved network efficiency. SpaceX cites deployments across construction, agriculture, retail, telecom, hospitality, aviation, maritime, and land mobility, including remote worksites and vehicles, and names customers such as @United Airlines, @Carnival, @Maersk, and @John Deere, adding that enterprise engagements often start as backup connectivity and then become primary, with no enterprise customer over $750,000 annual revenue voluntarily discontinuing since 2023. For government and national security, it highlights #Starshield, built on the same infrastructure as Starlink, offering missions like Earth observation and global secure communications with high-assurance cryptographic capabilities, positioned as a source of long-term contract revenue. The filing also emphasizes scaling via channel partners in select geographies and deepening telecom integration through #StarlinkMobile partnerships with about 30 mobile network operators across six continents as a complement to terrestrial cellular infrastructure.
22. College Students Are Rapidly Losing the Ability to Read
Literature and writing instructor Tyler Jagt argues there is a measurable, generational collapse in #sustained reading and writing, visible in his classroom where students could not finish an assigned 20-page article and reported losing track of what it was about. He points to 2024 #NAEP results showing 12th grade reading scores at their lowest since 1992, with nearly a third of students below the “basic” level, and to an @Annie E. Casey Foundation report finding 70 percent of fourth graders are not proficient readers. Jagt links the trend to increasing reliance on #generativeAI in higher education, where students use tools like #ChatGPT to summarize or complete work, and he notes research raising doubts about educational benefits, including an @MIT study in which AI-assisted writers showed lower brain activity in creativity-related areas and 83 percent could not quote a line from essays they had just produced. He also cites evidence that #smartphones sap attention, including a 2017 study finding that merely having a phone nearby reduces cognitive capacity, framing students’ difficulty sustaining attention as potentially tied to atrophying neural pathways. Overall, he contends academia is responding with ad hoc coping and exhaustion rather than the structural overhaul needed to rebuild reading endurance and attention.
@AMD released its first official, estimated benchmarks for upcoming EPYC “Venice” chips using #Zen6, claiming the 256-core flagship delivers 3.3x the rack-level performance of @Nvidia Vera under a fixed 100kW power budget. The comparison is based on a modeled rack deployment, where AMD estimates node counts from CPU TDP plus additional component power, then multiplies by measured single-node results across benchmarks including SPEC CPU 2017 integer throughput, SPECjbb 2015, NGINX load (WRK), Redis, Memcached, and MySQL TPROC-C. Because AMD did not test Vera directly, it estimates Vera by scaling @Nvidia Grace results by 1.63x from Phoronix-published Vera data, and it estimates the 256-core Venice by applying a 1.7x scaling factor over EPYC 9965 plus internal testing, noting the figures are only a directional comparison. The article cautions that simple linear scaling can break down at rack scale due to interconnect, thermal, and power constraints, and frames the results as a response to Nvidia-curated Vera results ahead of AMD’s upcoming Advancing AI event where more Venice and enterprise roadmap details are expected.
@Google reportedly ordered @Intel to package more than 3 million of its TPU accelerators in 2028 after months of testing Intel’s advanced packaging, while @Nvidia is said to be evaluating Intel for a future multi-die GPU tied to its 2028 Feynman architecture and @SK hynix is testing HBM reliability with Intel packaging. The push is driven by a severe #CoWoS bottleneck at @TSMC, whose CEO C.C. Wei said capacity will take a long time to meet demand, with advanced-node capacity running about three times short and CoWoS effectively sold out for years; much of current demand is concentrated with Nvidia, plus Broadcom and AMD. The article frames #EMIB as the only credible second-source packaging option that AI chip makers can qualify at volume before the end of the decade, and notes earlier reports that Google and Amazon were in discussions with Intel now appear to be solidifying into unit counts, timelines, and memory-qualification steps. It contrasts CoWoS using a large silicon interposer with EMIB using small embedded silicon bridges in an organic substrate, citing higher package utilization for EMIB and Bernstein estimates suggesting lower per-chip packaging cost than CoWoS, though with limited external production track record. It also notes a key tradeoff that standard EMIB has longer, resistive power paths that may be insufficient for high-current HBM4-class accelerators, motivating variants like #EMIB-T that add through-silicon vias for improved power delivery.
Samsung Heavy Industries is moving to commercialize a 50MW floating #AI #data center platform through agreements with Greek shipowner Capital Clean Energy Carriers and @Lloyd’s Register, alongside a joint effort with @Supermicro to validate running AI servers at sea. The three way memorandum assigns Samsung Heavy technology and construction, Capital project sourcing and investment, and Lloyd’s Register regulation and certification, and the design combines seawater cooling with onboard power from #solid oxide fuel cells fueled by #LNG, with optional shore power via subsea cables when moored. The project received approval in principle from the American Bureau of Shipping and Lloyd’s Register, and is positioned as a way to bypass long grid connection queues affecting land based builds in the U.S. and Europe, while offering mobility and flexible siting. A key technical risk is hardware durability under vibration, tilt, salt, and humidity, with Samsung Heavy developing positioning and sealing technologies and Supermicro validating server operating conditions in river and marine environments. The article notes growing activity in floating and subsea data centers, and mentions a letter of intent with @OpenAI as a potential anchor tenant, but says there is not yet a signed customer contract for a named deployment.
Thoma Bravo founder @Orlando Bravo said fears of an AI-driven #SaaSpocalypse are over, arguing that #AI is now an enormous tailwind for software companies rather than an existential threat. He pointed to a February sell-off in #SaaS stocks after Anthropic unveiled advanced tools for its Claude co-working agent, but said investors underestimated software firms ability to evolve and incorporate new infrastructure. Bravo said Thoma Bravo portfolio companies are largely booming because of AI, and that about 50% of their new revenue comes from AI and #agentic tools, which he expects to merge with software into new agentic solutions for corporate customers over the next few years. He noted software stocks have rebounded, citing the iShares Expanded Tech-Software Sector ETF jumping 21% in May and rising more than 9% over three months, while cautioning that governance, cybersecurity, and return-on-investment questions around newer agentic tools still need to be worked through.
27. Diabetes org apologizes for ejecting scientists over criticism of Trump
Under intense backlash, @American Diabetes Association CEO @Charles Henderson apologized after the ADA forcefully removed five prominent diabetes scientists from its annual meeting for distributing an ADA-journal editorial criticizing the @Trump administration’s harm to #biomedical research. The scientists, including @Steven Kahn (editor in chief of Diabetes Care) and former ADA president @Desmond Schatz, were handing out copies outside the opening session where @Jay Bhattacharya was scheduled to speak before canceling, with Rick Woychik speaking instead, and police in New Orleans escorted them out, confiscated badges, threatened arrest, and the ADA barred them from the rest of the conference. Henderson’s video apology contrasted with earlier ADA statements that framed the ejection as a #code of conduct issue, a lack of prior approval to distribute materials, and compliance with #501c3 rules requiring a nonpartisan environment. Critics noted those rules do not bar leaders from expressing personal political views or addressing public policy issues, and the incident spread on social media, spiking traffic to the editorial, prompting resignations, and drawing a letter from over 40 ADA officials calling the decision outrageous and its justifications unpersuasive. The apology seeks to repair the rift and “bring our community back together” after the organization’s handling of the protest drew widespread condemnation within the diabetes research community.
28. US lab discovers new way to build powerful magnets without rare earths
U.S. scientists developed new #AI tools to help identify suitable materials for making #permanent magnets without using #rare earth elements. The article states that these tools support the discovery of the right materials needed to replace rare earths in magnet production. This suggests a computational, materials selection approach aimed at enabling strong magnets while reducing reliance on rare earth supply chains. By improving how candidate magnet materials are found and evaluated, the work targets a pathway to powerful rare-earth-free magnets. The article frames the advance as a new method centered on AI-assisted materials discovery for rare-earth-free permanent magnets.
29. Angry bug hunter with Microsoft beef drops new Windows 0-day
Security researcher Nightmare Eclipse, who claims to be a disgruntled ex-@Microsoft employee, publicly disclosed a new Windows zero-day called RoguePlanet shortly after June Patch Tuesday. The flaw targets #MicrosoftDefender and is said to work on fully patched Windows 10 and Windows 11, with proof-of-concept exploit code released; if an attacker wins a race condition, it enables local privilege escalation to SYSTEM. The researcher alleges @Microsoft ignored reports and mistreated them, and RoguePlanet is described as the seventh Microsoft zero-day they have disclosed before a vendor fix, with several earlier releases reportedly exploited in the wild prior to patching. @Microsoft told The Register it is aware of the reported vulnerability and is actively investigating its validity and applicability, and reiterated its commitment to updating impacted products to protect customers.
The technology sector is projected to face continued layoffs through 2026, with over 150,000 job cuts already announced by major companies such as @Meta, LinkedIn, Salesforce, and Intuit. #Layoffs have surged due to shifting economic conditions and company restructuring aimed at reducing costs and refocusing business strategies. This wave of job losses reflects broader challenges within the tech industry, including slowing growth and market uncertainties. Analysts suggest these cutbacks could reshape workforce dynamics and prompt ongoing industry adjustments over the next few years. As firms streamline operations, the trend highlights the growing need for adaptability in an evolving tech landscape.
32. TSMC: World’s largest chipmaker does not rule out price rises as costs increase
The world’s largest chipmaker, #TSMC, told the BBC that inflation is increasing its costs and it does not rule out raising prices, which could ripple through #AI infrastructure and eventually consumer electronics. CFO Wendell Huang said any changes would not be sudden “fourfold, fivefold” jumps, arguing pricing reflects the company’s “technology leadership” and “manufacturing excellence,” while CEO @CC Wei said he would “like” to raise prices as competitors have done. Huang also rejected the idea that the #AI boom is a bubble and said TSMC’s expansion to the US, Germany, and Japan is driven by customer demand rather than geopolitical pressure from Washington or Beijing. Still, he said the most advanced chip production will remain in Taiwan, and that recreating Taiwan’s manufacturing ecosystem in the US could take five to 10 years or longer, contrasting with US industrial policy expectations tied to TSMC’s $165bn Arizona commitment. With AI-driven demand pushing TSMC to expand as fast as possible and markets debating whether AI infrastructure spending is sustainable, the company’s pricing and capacity choices sit at the center of global #semiconductor and #US-China tensions.
33. Meta reveals over 20,000 Instagram accounts hacked and stolen using AI support bot
@Meta reported that 20,225 Instagram accounts were compromised after attackers exploited a flaw tied to its AI-assisted High Touch Support (#HTS) account recovery flow to obtain password reset links. In a filing with the Office of the Maine Attorney General, @Meta said it found the bug on May 31, 2026: a separate code path failed to verify that the email entered for a reset matched the email on the account, so reset links could be sent to unassociated emails. @Meta said it has no evidence of #data exfiltration, but acknowledged attackers likely had access to information inside affected accounts, including contact details, birth date, posts, photos and videos, stories, direct messages, account activity, profile data, and linked services. In response, @Meta disabled #HTS, reset passwords for affected accounts, forced a security checkpoint with re-authentication, and said it will fix the authentication check before relaunching while also reviewing similar recovery flows across its platforms. A Huntress advisor warned that embedding #AI into support and identity workflows creates logical attack surfaces, and tools that can trigger privileged actions like password resets need rigorous verification and access controls.
34. How and Why to Fight Back Against Social Media Bans
Several U.S. states, including Massachusetts, Idaho, Minnesota, North Carolina, South Carolina, Illinois, and California, are advancing proposals that ban or heavily restrict young people’s access to social media, framed as #childrens online safety but described here as a shift toward mass censorship and surveillance. The text argues that earlier #age gating efforts that were presented as limited to adult content have expanded, because once government-backed infrastructure exists to collect and verify user data, it can be used to restrict broader categories of lawful speech and access to the “modern public square.” It describes how these bills work, ranging from outright bans like California’s AB 1709 to schemes that allow access only with additional data such as verifiable parental consent, and it notes feature controls like default privacy settings, time limits, or notification preferences for accounts that fail an age gate. It distinguishes two main mechanisms: #age verification, which can require collection of sensitive data such as government ID and biometrics and may rely on third-party verification as in Florida’s HB 3, and #behavioral age estimation, which uses existing platform data and algorithms or #AI to infer age as proposed in Minnesota’s HF 1438 and South Carolina’s H 4591. It contends that behavioral estimation does not eliminate risk because the underlying data is often gathered for targeted behavioral ads, and using imperfect datasets to decide access can entrench harmful outcomes.
Taiwan is considering much stricter #export controls that would restrict shipments of high-end #AI chips and related systems to all customers in China, not just blacklisted firms like Huawei, and would make diversion and smuggling prosecutable as a criminal offense. The proposal, discussed in the context of trade talks with the United States, would likely use a processing-power threshold similar to U.S. rules such as the Total Processing Performance cutoff in #ECCN 3A090, aligning with the approach behind the #SAFEChipsAct. Taiwan currently does not treat unauthorized AI chip exports to China as a standalone crime, and prosecutors have pursued suspected diversions through other charges, such as a May case in Keelung involving about 50 @Nvidia equipped servers detained on alleged document forgery. Because Taiwan manufactures most of the world’s AI servers through firms like Foxconn, Quanta, Wistron, Wiwynn, and Inventec, a threshold-based rule would directly target the export of assembled server systems that could otherwise be diverted downstream even if @TSMC is already barred from making advanced chips for Chinese customers. The report says Taipei has agreed to directionally follow U.S. curbs but has not finalized how far it will go, and any move may prompt a response from Beijing.
36. GM to Develop Sodium-Ion Battery Cells—for Energy Storage, Not EVs
General Motors says it will develop and build #sodium-ion battery cells with partner Peak Energy, but the cells are intended for stationary #battery energy storage systems, not near-term EV use. GM’s current EV lineup uses #NMCA cells from its Ultium Cells joint venture with #LG Energy Solution, the rebooted 2027 Chevrolet Bolt uses #LFP cells sourced from China’s #CATL, and it is also pursuing #LMR as a third lithium-based chemistry, making sodium-ion a fourth. The company argues sodium-ion will reduce lifetime costs for large, bunker or container style storage systems expected to last 20 to 25 years, potentially even versus LFP, helped by sodium’s low cost and abundance, precursor supply not being controlled by Chinese companies, and improved cold and heat tolerance that can eliminate the need for liquid cooling. In an interview, GM battery and sustainability VP @Kurt Kelty and engineer Andy Oury said the effort should also help GM lower EV battery costs faster by expanding its development capabilities and battery business. The plan targets growing demand for grid and data-center storage while positioning GM to benefit from cheaper, simpler, more robust cell technology and supply chains.
37. Microsoft patches YellowKey, GreenPlasma, MiniPlasma zero-days
@Microsoft shipped fixes in its June 2026 Patch Tuesday updates for three recently disclosed Windows zero-days that enable #privilege-escalation to SYSTEM or access to #BitLocker-protected data. The issues include GreenPlasma (CVE-2026-45586) in the Collaborative Translation Framework (CTFMON) and MiniPlasma (CVE-2020-17103) in the Cloud Files Mini Filter Driver, both allowing local attackers to obtain a SYSTEM shell on fully patched Windows systems, plus YellowKey (CVE-2026-45585), a WinRE backdoor that can let attackers with physical access bypass BitLocker on unpatched Windows 11 and Windows Server 2022/2025 devices. All three were disclosed last month by the researcher “Nightmare Eclipse” in protest of @Microsoft Security Response Center handling of disclosures, and @Microsoft published mitigations for YellowKey while criticizing the public proof-of-concept release as violating coordinated disclosure practices. The report places these patches in the context of additional recent leaks by the same researcher, including other LPE and #MicrosoftDefender-related zero-days and a newly disclosed “RoguePlanet” exploit, alongside @Microsoft’s shifting public stance from legal threats to saying it would involve law enforcement when laws are broken.
38. Brit workers waste nearly six hours a week ‘botsitting’
A report from the Work AI Institute, part of Glean Technologies, says UK workplace #AI adoption is widespread but productivity gains are being eroded by time spent “botsitting,” meaning hand-holding tools and fixing their mistakes. In a survey of 1,500 digital workers for “The Work AI Index: UK 2026,” 90 percent say they are required to use AI, 80 percent use multiple AI tools weekly, and 39 percent use four or more, yet only 18 percent say AI has significantly improved organizational performance. Respondents estimate AI automation saves about 12 hours a week, but they also waste about 5.8 hours weekly making outputs usable, and report that 36 percent of AI sessions fail outright, requiring restarts or substantial rework, with roughly an hour of cleanup for every hour spent getting AI output. The wasted time goes into reloading context, supervising results, catching hallucinations, verifying outputs that look plausible, and iterative re-prompting or switching models, effectively turning workers into an integration layer that #APIs and the #ModelContextProtocol do not fully solve. The report says fatigue leads to lower diligence, with 70 percent admitting they pass on the first output that seems “good enough,” which can shift error-fixing onto colleagues.
39. Prostate beam therapy spares patients daily hospital trips
A new form of #prostate cancer treatment using beam therapy reduces the need for daily hospital visits, improving patient convenience and recovery. This innovative approach targets tumors with precision, minimizing damage to surrounding tissues and lessening side effects. Clinical trials indicate that patients undergoing this therapy experience fewer disruptions to their daily lives while maintaining effective cancer control. The reduced treatment schedule also lowers the strain on healthcare facilities and resources. This advancement marks a significant step in cancer care, enhancing patient quality of life and treatment efficiency.
That’s all for today’s digest for 2026/06/11! We picked, and processed 38 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! 🚀
