↵ select ↓ ↑ navigate esc close

Daily Briefing: October 27, 2026, evening

Daily Briefing ·

Listen: https://blossom.buildtall.systems/ee828f1645ab9fc79a304c74321f4bd5e3cf9483cb0c3a7b3c7cd74a3b0aa081.mp3


The briefing for October 27, 2026 covers fifty-eight articles on Anthropic's initial public offering and the sharpening AI safety controversy, the AI compute market and China's stock correction, new world models and office software rivals, PHP vulnerabilities and a broad run of breaches, scams, and ransomware enforcement, Internet disruptions, Offline Protocol SDK releases, politics in Israel, Lebanon, and Serbia, Nautilus science features, Apple and American manufacturing news, and essays on influence, grief, memory, and environmental accountability.

The first theme is Anthropic and the AI safety debate. Five articles converge on the company's coming public offering, its internal contradictions, and disputed questions about AI regulation and capability audits.

Anthropic S-1, The Cost of Compute, Who Is Paying? from Stratechery reports that Anthropic's initial public offering is finally here. The piece summarizes the situation as one in which the core business is very profitable, but competition may eat all of the profits, even as investors are set up for endless dilution.

The long analysis Anthropic Versus Anthropic from Don't Worry About the Vase traces the company's split identity. The article opens by calling Anthropic frustrating, a company that sometimes takes bold stands, builds a unique governance structure, and maintains a relatively strong though inadequate safety culture, yet at other times does deeply disappointing things. The piece revisits the leaked draft bill that would have preempted significant state law, reporting on a direct response from Dario Amodei to critic Nathan Calvin. Amodei lists costly safety-adjacent actions including involvement with SB 53 in California, aggressive biological safeguards on models, delaying the release of Mythos-class models to strengthen cyber safeguards, advocating for government power to block model releases, taking a public stance in February 2026 that could have damaged the company, supporting export controls, restricting sales to PRC-controlled companies, extensive model cards and risk reports, and the new commitment of five percent of company shares to a public benefit fund. The article then recounts Nathan Calvin's reply, which restates four key claims: that Anthropic is racing to build recursively self-improving superintelligence that by its own lights could kill billions, that this risk is very hard to justify, that Anthropic often works to keep governments from tying its hands, and that even if the AEGIS Act was an honest mistake, that itself shows a lack of gravity. The piece draws in Kelsey Piper's reporting from inside Anthropic, where Piper says there is no plan for alignment, no plan to avoid concentration of power or loss of control, and no plan for what a good world with superintelligence would look like. It criticizes Anthropic's Public Benefit Report for formatting that seems designed to be misread as saying the technical problems of AI safety are solved, when the smaller print later says alignment is far from solved. The piece then covers Anthropic's agreement with METR to provide employee-like access to a small siloed team, with METR's commitment to publish analysis without veto, and reports METR's disagreement with the methodology of Anthropic's August 2026 risk report while still affirming that overall risk is low. It notes METR's specific worry that offline monitors would have missed a Mythos 5 supply chain attack, and that redacted material, especially a footnote with examples of historical concerning model behavior, increased METR's worry about sandbagging and targeted undermining of safety-relevant research and development. The conclusion returns to the internal split: policy staff have gone rogue in favor of business interests, the culture is insular, and the costly sacrifices made on biosecurity and China do not reassure about alignment, takeover, and disempowerment.

NYT debate on the rules for AI companies from Marginal REVOLUTION reports that Emily Bazelon moderated a five-person symposium in the New York Times magazine. The post quotes an exchange between Tyler Cowen and an interlocutor named Bores, in which Cowen says federal laws preempt state lawmaking in many other areas and that courts have upheld it many times. When Bores asks whether preempting states is pro-democracy in giving people a say, Cowen responds that it is pro-American Constitution. The post notes that most of the panel disagreed with most of the author's points.

Hiding Capabilities in AI Models from Schneier on Security links to research titled "AI Models Can Provably Hide Arbitrary Capabilities." The abstract describes backdoor attacks that embed encrypted neural networks within host model weights, decrypting and executing them only upon receiving a secret trigger. Using digital lockers, these hidden circuits remain provably hard to elicit and interpret under cryptographic assumptions, even with full access to model weights. The researchers implement proof-of-concept constructions in PyTorch using a small chemical reaction prediction task as a proxy for CBRN-relevant capabilities. The results demonstrate resistance to fine-tuning, weight noising, gradient-based trigger search, and latent adversarial training, and the paper concludes with a limit of elicitation-based capability audits.

Frontier AI labs say their models refuted the Exponential Time Hypothesis. What does that mean? from TechCrunch reports that TechCrunch spoke with the MIT computer scientist charged with explaining OpenAI's latest mind-bending discovery. The article concerns claims from frontier AI labs that their models refuted the Exponential Time Hypothesis.

The next theme is AI compute, capital, and the China correction. Four articles cover new inference chips, a large series A, a burst bubble, and a ten-billion-dollar fund.

Nvidia backs Indian startup helping companies run AI models with fewer GPUs from TechCrunch reports that Simplismart says its annualized revenue run rate has grown fourteen-fold in six months, with the United States accounting for more than half its revenue. The startup helps companies run AI models with fewer GPUs and has Nvidia backing.

General Compute raises $100 million Series A, countering the neocloud disaster narrative from TechCrunch reports that General Compute is betting on inference chips from outside the Nvidia ecosystem.

In China, the AI Bubble Has Already Burst from Slashdot quotes a Reuters report. The mainland STAR50 index of AI and semiconductor leaders tumbled forty percent in less than four months, vaporizing some four hundred eighty billion dollars in market value. The blue-chip index dropped about twenty percent from June highs. A Beijing fund manager named Dong Baozhen says the stock market evolved into a speculative crazy bull and now the bubble is bursting. Goldman Sachs reports AI stocks account for forty-two percent of China's market capitalization and fifty-seven percent of trading turnover. Hong Hao of Lotus Asset Management says the downturn will stop many companies trying to get listed, while Leo Wang of Qianchuang Capital says overemphasizing technology content in major indexes created a severe bubble in certain sectors and a serious drain on others. Fund manager Wang Zhuo argues that Chinese chipmakers and AI suppliers will eventually end up in a price war, and that all bubbles inevitably burst.

Disruptive raises $10B fund to make big bets on at least 10 companies from TechCrunch reports that Disruptive has made headlines in recent years for pouring money into companies like chip startup Groq and the AI voice-generation startup ElevenLabs.

The next theme is AI products, world models, and browser integration. Three articles cover a browser standard and two reports of Google's new world model.

Opera introduces a new web standard to let users deploy AI locally in their browser from TechCrunch reports that the new standard, WebAI, is designed to work with a variety of AI models.

Google's Genie 4 world model adds audio, more realistic worlds, and real-time steering from TechCrunch reports that Google DeepMind released Genie 4, a world model with real-time audio, more realistic simulations, and mid-simulation text prompts, in research preview for trusted testers.

Google's Genie 4 World Model Adds Audio, More Realistic Worlds from Slashdot reprises the TechCrunch report. Genie 4 generates the most photorealistic and interactive worlds yet, allowing users to type prompts mid-simulation to change the scene. The model generates real-time audio matching the simulated world, including music, ambient noises, and interactive dialogue. It allows freeform edits like changing weather or camera angle in real time and adding new people or objects. It maintains a consistent scene for up to two minutes, double its predecessor's limit, with better physics and more human-like characters. Tejas Kulkarni, research director at Google, says the future will see hybrid world models where Gemini writes code and text while Genie ingests that in real time and simulates reality at different levels of abstraction. Genie 4 is available to select testers, with a broader rollout planned for Google AI Ultra subscribers.

The next theme is the race for the office suite. Two articles cover Denmark's failed open-source pilot and a Meta director's new startup aimed at Microsoft Office and Google Workspace.

Denmark Gives Up On Replacing Microsoft Office With Collabora from Slashdot reports via Tom's Hardware and the Danish publication Computerworld that Denmark's Ministry of Digitization abandoned a pilot to replace Microsoft Office with the open-source Collabora suite less than six months after it began. The pilot involved thirty-eight employees and was initiated by then Digitization Minister Caroline Stage to reduce dependence on American technology. By the time the government called an early election in February 2026, the experiment had reportedly ended, with employees complaining about issues with style, paragraph formatting, blank pages, and challenges with tracking changes. Professor Mikkel Flyberbom of Copenhagen Business School complains in an email that the ministry never took the opportunity to learn from the pilot and share those experiences with other parts of the public sector.

Meta AI director Rohit Patel leaves to take on Microsoft Office and Google Workspace from TechCrunch reports that Rohit Patel is departing Meta after eight years working on AI evaluations and reinforcement learning. His startup, Ndoc, aims to take on Microsoft Office Suite.

The next theme is PHP vulnerabilities and the capture the flag challenge. Three articles cover a week of certificate compromise and PHP flaws, a government warning, and a Cloudflare contest.

Introducing the Cloudflare PHP Capture the Flag challenge from the Cloudflare Blog reports that last week was one for the Internet security history books. First, eight eMudhra intermediate certificates were compromised, creating the possibility of forged certificates to impersonate servers. Then came php2shell, serious vulnerabilities in PHP, the runtime behind a large portion of the Web. Cloudflare shipped client updates and multiple mitigations, updated the web application firewall with new rules, and introduced full transaction detection, all available on all plans including Free. The post announces the Cloudflare PHP Capture the Flag challenge, run through its Bug Bounty program on HackerOne. A deliberately vulnerable PHP installation running stock PHP 8.5.11, Apache 2 via mod_php, and an empty index.php inside a Cloudflare Container sits behind the default WAF protections, with a file called /flag inside the sandbox. Security researchers must bypass the WAF protections, exploit the PHP sandbox, and recover the contents of /flag, then verify through the CTF interface and submit a private report. The challenge focuses on CVE-2026-109618 and CVE-2026-109619, the two PHP vulnerabilities behind php2shell. File disclosure without remote code execution, remote code execution without flag recovery, local-only exploits, and rediscovery of the known PHP vulnerabilities are not eligible. The blog stresses that WAF protection is not a replacement for patching.

CISA warns of PHP bug exploitation attempts, gives federal agencies days to patch from The Record reports that CISA added the widely publicized PHP vulnerabilities to the Known Exploited Vulnerabilities catalog, but did not specify what kind of exploitation it was observing.

US government says hackers are now exploiting critical PHP web bugs from TechCrunch reports that the U.S. government cyber agency warns hackers have figured out a way to exploit the popular PHP software, which is used in seventy percent of the web's software.

The next theme is breaches and institutional cyber incidents. Four articles cover intelligence sharing, a Japanese provider breach, a hospital diversion, and repeated breaches of police data suppliers.

Australia's spies are sharing intelligence with hacking targets — and want them to share back from The Record reports that more than half of the individual warnings from the Australian Signals Directorate drew on previously secret material that had been cleared for disclosure, underscoring the growing role of intelligence sharing to prevent attacks.

Japan internet giant GMO says hackers posted stolen customer data from TechCrunch reports that Japan's GMO Internet is verifying alleged customer data posted by suspected hackers following a cloud service breach.

University of New Mexico hospital diverts ambulances after cyberattack from The Record reports that for about two hours on Monday, the organization diverted ambulances to health systems in Albuquerque and Santa Fe, and suspended outpatient surgeries and appointments for the remainder of the day.

Hackers repeatedly breached suppliers holding UK police data, watchdog finds from The Record reports that His Majesty's Inspectorate of Constabulary and Fire & Rescue Services says British police data has been breached several times this decade through third-party IT suppliers.

The next theme is malware, scams, and ransomware enforcement. Four articles report malicious apps, a ransomware recovery fraud indictment, an international scam operation, and a sentencing.

Malicious apps prey on Russians' anxiety about fuel prices, drone strikes from The Record reports that cybercriminals are stuffing infostealer malware into bogus apps aimed at tracking gas prices, police activity, drone strikes, and other information of concern to average Russians.

Cybersecurity CEO Indicted For Allegedly Scamming Millions From Ransomware Victims from Slashdot quotes a ProPublica report and the Justice Department. Florida man Zohar Pinhasi, known by the alias Zack Green, told customers he could recover encrypted files without paying hackers when in reality he paid the ransom and charged victims a hefty fee on top. Federal investigators found that over the course of the scheme, Pinhasi allegedly charged clients more than nineteen million dollars and paid more than eight million dollars in ransom payments. In one example, Pinhasi made a ransom payment of about eight thousand two hundred dollars but charged the client about one hundred fifty thousand dollars. U.S. Attorney Joseph Nocella says Pinhasi re-victimized his clients while extracting a hefty profit, and the indictment mirrors ProPublica's 2019 findings about MonsterCloud. If convicted, Pinhasi faces up to twenty years in prison.

US indicts Malaysian national after Madagascar scam operation uncovered from The Record reports that the Justice Department action against five foreign nationals, plus a sting in Madagascar, highlights the expanding global reach of the scam center industry.

Nefilim, Hive ransomware operator given five-year-plus sentence after guilty plea from The Record reports that Ukrainian national Artem Stryzhak, thirty-six, had pleaded guilty last year to charges related to participating in at least two well-known ransomware groups. He has now been given a sentence of more than five years.

The next theme is Internet disruptions. Two articles cover quarterly disruptions and new restrictions in Moscow.

What took the Internet offline in Q3 2026 from the Cloudflare Blog reviews disruptions observed by Cloudflare Radar between July and September. The post records forty-nine disruptions in total: government-directed shutdowns accounted for nineteen incidents, including eighteen exam shutdowns in Iraq; power outages followed with fourteen incidents; and natural disasters accounted for five. In Cuba, four power-related outages saw traffic volume decreases ranging from eighteen to fifty-nine percent, including a three-day blackout from September 18 to 20. In Honduras, a power failure at an interconnection facility on September 10 affected at least eleven major providers and cut HTTP traffic across the country by sixty-three percent over roughly three hours. In Ukraine's Sevastopol, a July 5 power outage caused an eight-hour blackout with HTTP traffic dropping ninety-eight percent, and an August 12 drone strike on energy infrastructure caused a ten-hour disruption with up to ninety-five percent decline. A cooling system failure at a US data center in Phoenix caused a near-total drop in Internet traffic from Namecheap on August 13. Power-related disruptions also appeared in Eswatini, Botswana, Tajikistan, Kenya, Saint Lucia, Georgia, and Gabon. Natural disasters included Super Typhoon Bavi in Guam, a magnitude seven point one earthquake in Kumamoto, Japan, a magnitude seven point four earthquake in Colombia, and flash floods in Nepal. Iraq conducted scheduled exam shutdowns throughout the quarter, each lasting fifteen to ninety minutes. Ethiopia's Tigray region experienced a ten-day government-directed shutdown with traffic dropping by one hundred percent. Russia's Far East saw fiber-optic cable damage during elections affecting Chukotka, Magadan Oblast, Kamchatka Krai, and Sakhalin Oblast. The post concludes that the quarter was shaped by shutdowns and power outages, but also by resilience, with fourteen disruptions recovering in under an hour and sixteen in under three hours.

Moscow faces internet disruptions as foreign leaders gather for summits from The Record reports that restrictions will be in place from October 27 to 29, coinciding with the Gas Exporting Countries Forum summit and the Russia-Africa summit. Russia's Ministry of Digital Development announced the measures Monday.

The next theme is privacy and surveillance. Two articles cover a surveillance company CEO's claims and a guide to low-data travel.

It's not the bots, Flock from 404 Media reports on leaked audio from a private police conference in Orlando. Flock CEO Garrett Langley told an audience of police chiefs that anti-Flock protests around the country are not real. He said his wife advised him never to apologize, said he would prefer never to do interviews again, and claimed without evidence that social media opposition to Flock is driven mostly by Russian bots. Langley said social media chatter on Flock is down about ninety-five percent from July and that it went down fifty percent last week when Ukraine bombed a Russian data center with a drone, repeating a disproven theory about bot activity. He called the protests the same people with the same signs and the same message, not organic. The article notes Langley recently lied about Flock's role in tracking a woman who self-administered an abortion, calling 404 Media's coverage entirely false.

How to protect your privacy while traveling from the Proton Blog explains that travel creates a trail of personal data through smartphones, tickets, card payments, and cameras. The article walks through leaving the smartphone behind and keeping it in a Faraday bag, using cash and paper tickets where possible, printed timetables and paper maps for transport, calling properties directly and asking what information they require for accommodation, accepting that CCTV and live facial recognition are hard to avoid, recognizing that driving creates records through automated number plate recognition and connected vehicles, and using Tails and Tor on public computers. Each workaround comes with trade-offs; cash does not erase the record of an ATM withdrawal, paper tickets do not hide cameras, and signing into personal accounts over Tor identifies the user to those services. The piece concludes that privacy means choosing trade-offs: reducing unnecessary exposure rather than assuming every trace can be erased.

The next theme is the Offline Protocol SDK release. Two articles share the same release number and component.

SDK 0.31: a Ruby gem, Flutter parity, live BLE signal and sessions that heal from the Offline Protocol blog reports that SDK 0.31.0 publishes a Ruby binding to RubyGems, gives the Flutter plugin the React Native module's full native surface, reports live Bluetooth LE RSSI as a peer_rssi event, and lets devices that share only a Nostr relay form a session, with split sessions healing themselves.

SDK 0.31: a browser runs a full Offline Protocol node from the Offline Protocol blog reports that SDK 0.31.0 runs the full engine, MLS included, in a browser worker with durable storage. Browsers link over WebRTC, find each other on one network, reach any device over the application's Nostr relays, and link to a phone or a board over Web Bluetooth.

The next theme is political transition in conflict-affected states. Three New Lines Magazine pieces cover right-wing youth in Israel, Amal's role in Lebanon, and Serbia's new democrats.

Why the Majority of Israel's Youngest Voters Identify as Right Wing from New Lines Magazine reports that the Israeli eighteen-year-olds voting for the first time today were babies when Benjamin Netanyahu was elected prime minister in 2009. Three-quarters of eighteen- to twenty-two-year-olds identify as right-wing or far-right, according to a Channel 12 News poll. The article traces the shift to the Second Intifada, the COVID-19 pandemic, the October 7 Hamas-led attack, the war in Gaza, and the normalization of the extreme right, including Itamar Ben-Gvir and Bezalel Smotrich, now respectively minister of finance and minister of security. It describes Lior Schleien's one-man civil society initiative to rescue Israeli youth from what he calls incipient fascism, holding town hall meetings with high school students in videos posted online. The teenagers assure him they are open to changing their minds but few seem convinced; one angrily asks whether it is better for Israel to be democratic than Jewish, while another says he wants to kill or expel all Arabs living under Israeli sovereignty. The piece notes Schleien does not explain the contradiction between arguing for the democratic rights of Palestinian citizens, who make up twenty percent of the population, and preserving Israel as a Jewish state. It concludes that we will soon see how those attitudes affect the results of Tuesday's election.

Can Amal Help Lebanon Find a Way Out of War? from New Lines Magazine opens with Speaker of Parliament Nabih Berri's homage to the vanished cleric Musa al-Sadr on the forty-eighth anniversary of his disappearance, and his warning against negotiating under fire. The article examines the Amal Movement's history from Sadr's founding of the Movement of the Deprived in 1974, through the Lebanese Civil War, the War of Brothers with Hezbollah, and the division of labor that left Amal embedded in state institutions while Hezbollah kept its weapons. It reports Amal officials' skepticism about the current framework agreement with Israel, quoting one political bureau member who says Israel has not given up anything and that the country would need decades for Israel to withdraw from just a bit of Lebanese territory. Amal supports indirect negotiations leading to Israeli withdrawal to the international border and Lebanese army deployment south of the Litani, plus bringing Hezbollah's weapons under state control. The party stayed inside Prime Minister Nawaf Salam's government despite opposing the March Cabinet decision to prohibit Hezbollah's armed activities. The piece describes Amal's ties to both Washington and Tehran, including a representative office in Tehran and a Presidency Council member responsible for relations with Iran, and its bet on reviving the Pakistan agreement from April. It concludes that if Lebanon is to find a way out of its political and security crises, it is difficult to see how Amal can be left out.

Serbia's Victorious New Democrats Face an Old Question from New Lines Magazine reports that for the second time in just over a quarter-century, Serbia has thrown off the yoke of one-man rule. Serbia ousted Aleksandar Vučić and his ruling Serbian Progressive Party after their nearly fourteen-year tenure, following two years of protest that began with the Novi Sad railway station canopy collapse in November 2024, which killed sixteen people. The Student List parliamentary coalition won fifty-three percent of the popular vote and secured around one hundred forty seats in the two-hundred-fifty-seat National Assembly. Domestically the Student List ran on an anti-corruption platform, but its foreign policy agenda is virtually indistinguishable from that of the SNS because Serbian nationalism remains the most potent political force. The article argues that Serbia's regional ambitions and democratization are at odds, because pressing expansionist claims against Kosovo, Montenegro, and Bosnia has required security services, paramilitary and criminal proxies, and captured media that have also been turned on Serbia's own citizens. It points to Dejan Šoškić, former head of the central bank and possible next prime minister, who has suggested lustration of Vučić's network as a priority. The article argues lustration was precisely what was missing after Milošević's fall, and that the new government could declassify security archives and expose the truth about the wars of the 1990s and the assassination of Zoran Đinđić in 2003. It concludes that a government that recognizes Serbia's democracy depends on the rejection of its imperial machinations can begin to chart that course today.

The next theme is Nautilus science and nature. Five articles cover a national park, radar-baffling bats, spider sleep, memory, and a global food pyramid.

Take a Look at the Country's Newest National Park from Nautilus reports that the twelve-thousand-acre Chiricahua area in southeast Arizona just became the sixty-fourth national park in the U.S. Known as the wonderland of rocks, Chiricahua was shaped by ancient volcanic forces and contains dramatic geological features. It also has a rich human history, including the Chiricahua Apache, African-American soldiers stationed there in the 1880s, and homesteading families from the eastern U.S. Massai Point, named after a Chiricahua Apache man born in the Mescal Mountains, features spires and columns carved into rhyolite rock by millions of years of erosion.

When Texas Bat "Cities" Look Like Thunderstorms from Nautilus reports that weather apps generally modify their default settings to remove such scenes from Doppler radar views; otherwise live-feed watchers would think a torrential downpour is imminent.

How Jumping Spiders Sleep from Nautilus reports that jumping spiders sleep remarkably like humans, with cycles of wakefulness and REM.

The Teenage Girl Who Remembers Everything from Nautilus reports on a conversation with a scientist about a sixteen-year-old who has highly superior autobiographical memory.

There's a New, International Food Pyramid in Town from Nautilus reports that a new international food pyramid is based on research into dietary guidelines from around the world.

The next theme is Apple updates and foldable troubles. Four articles cover iOS, the M6 MacBook Pro announced twice, and a folding screen flaw.

Apple rolls out iOS 27.2 with new Siri AI languages and security fixes from TechCrunch reports that the latest Apple software update for iPhones and iPads includes new features and important security updates.

Apple unveils new 14-inch MacBook Pro with M6 chip from TechCrunch reports that the new MacBook Pro comes with twenty-four hours of battery life and now features support for Wi-Fi 7 and Bluetooth 6 connectivity.

Apple Updates MacBook Pro With M6 Chips from Slashdot reports via The Verge that Apple refreshed its standard fourteen-inch MacBook Pro with the new M6 processor. The M6 is Apple's first 2nm chip, bumping to a twelve-core CPU from ten cores in the M5, and offering up to 1.2 times faster multithreaded performance. The M6 also features a Dual sixteen-core Neural Engine for on-device AI tasks and an upgraded twelve-core GPU with a Neural Accelerator in each core. The laptop has three Thunder 4 USB-C ports that support charging and DisplayPort, a MagSafe 3 port, full-size HDMI, and an SDXC card slot. Pricing starts at one thousand nine hundred ninety-nine dollars with sixteen gigabytes of memory, the same as last year's M5 model. Preorders start October 28 and shipping begins November 6. The article also notes a touchscreen MacBook Pro is expected to debut tomorrow.

Apple's Foldable iPhone Duo Draws Early Complaints of the Screen Cracking from Slashdot reports via Bloomberg and iFixit that Apple has acknowledged an inner-display defect affecting a small number of its new one thousand nine hundred ninety-nine dollar iPhone Duo foldables after customers reported hearing a crunch while unfolding the device before cracks, lines, and other display failures appeared along the crease. Apple is replacing affected phones and investigating the problem. iFixit believes a failing hinge component may be pressing into and cracking the display rather than the panel being defective. Apple said it is aware of an issue impacting the inner display on a small number of iPhone Duo units and that most units aren't affected.

The next theme is reindustrializing American hardware. Five articles cover Anduril's laptop, the founder's view on the California wealth tax, an EV with vehicle-to-everything capability, and a manufacturing accelerator.

Anduril, Qualcomm launch BlueChip, a new kind of US-made laptop from TechCrunch reports that Anduril founder Palmer Luckey is vowing to bring computer manufacturing back to the U.S., and by that he doesn't just mean assembly, but every individual component, some of which are exclusively made in China today.

Anduril Partners With Qualcomm to Build Laptops On US Soil from Slashdot reprises the BlueChip announcement. The computer will use a Qualcomm Snapdragon SoC. Luckey says in a YouTube video that other companies might design their computers in the U.S., but Anduril makes them here, with parts made out of metals the U.S. knows how to extract, refine, process, and mill at scale at a reasonable price, and that goes for the entire laptop including the battery. The project will take a phased approach toward an end-to-end American supply chain, with the first device expected in April 2027.

Billionaire Palmer Luckey isn't leaving California over the wealth tax from TechCrunch quotes the native Californian saying the wealth tax is causing a lot of people to run who he doesn't want to be running, but he is planning on staying in California and going to keep fighting it.

Ford Fathom will be the cheapest American EV capable of powering a house from TechCrunch reports that Ford is rolling out two new chargers capable of powering a home or the grid, so-called V2x capability.

Re:Build, which spun out Slate Auto, raises $2.1B from TechCrunch reports that the manufacturing accelerator bills itself as America's answer to Shenzhen.

The next theme is personal growth and influence. Two short essays address disappointment and mimetic desire.

The half-life of disappointment from Seth's Blog asks how long it takes to move on when things don't turn out the way you pictured. The piece argues that reducing this window of disappointment unlocks the path to creativity and leadership. People who get over disappointment more quickly aren't less excited; they realize that getting attached to an outcome doesn't increase its likelihood, and in fact makes it less likely that the next project will be as bold. This is a common practice among professionals in every field.

Your Decisions Are Probably Not Your Own from The Next Step starts from a conversation about influence and argues that people are all more easily influenced than they believe. It connects this to mimetic desire, the idea that people don't actually know what they like and look at what others like. The piece suggests evaluating past decisions by lining them up with decisions made by friends, coworkers, and people followed online; recognizing which areas of life are most easily influenced; and surrounding oneself with people one wants to become. It stresses that influence isn't necessarily bad, but influence without awareness can be dangerous. The closing prompt asks what one area of life will see different decisions after evaluating how and by whom one is influenced.

The remaining articles stand alone.

Industry Executives Knew For Decades That Microplastics Were Infiltrating the Human Body from Slashdot quotes Susanne Rust of the Los Angeles Times. Internal memos among company scientists and executives reveal that for at least half a century, the plastics industry has been aware of and concerned about microplastic contamination in the human body. The concern goes back to 1974 and a scientific meeting at a New York City hotel, where a German gastroenterologist presented a paper showing for the first time that ingested particles like microplastics could infiltrate the rest of the body. His presentation was largely overlooked by government scientists but became a point of worried discussion among company scientists. Tracey Woodruff, professor of epidemiology and population health at Stanford University, who reviewed the documents, likened the silence to other industries that knew about pollutants like greenhouse gases, asbestos, and tobacco. Craig Davis, senior director of plastics chemistry at the American Chemistry Council, disputed that characterization, saying scientific debate over what the evidence shows is not silence, and that the industry is committed to advancing microplastics science by supporting independent research and policies.

Roundup #90: In which Noah joins a Japanese VC firm from Noahpinion announces that Noah Smith is joining the Japanese VC firm Coral Capital as a Venture Partner, a side gig consisting of a few blog posts and helping to introduce the firm to folks in the U.S. The reason is to encourage more American investment into Japan, the subject of his book The Weeb Economy, and to contribute to Japan's ability to defend itself. Coral's announcement quotes the firm saying the world's focus is shifting from bits back to atoms, and that reindustrialization, defense, energy, robotics, and physical AI are the defining investment themes of the next decade, playing directly to Japan's strengths. It says Japan is America's most important ally, the largest foreign direct investor in the United States, and the firm intends to be the one that bridges the corridor. The roundup then turns to several topics. First, it argues the humanities are dying, showing a chart from a study by Ober et al. and noting the humanities decline has been ongoing for over a decade while social science decline is post-pandemic; the article suspects difficulty of getting a job as a humanities PhD is the bigger factor. Second, it argues Paul Krugman's data center crowd-out story is fishy: the article says a positive shock to demand for data centers should only partially crowd out other spending, and the timing is odd because the big rise in real interest rates happened in 2022-23 while non-data-center construction fell off a cliff in mid-2025. Third, it reports China's local governments are in big trouble because property sales are crashing and average monthly debt servicing surpassed fifty percent of monthly income by the end of 2025 for all but seven Chinese provincial units, according to a paper by Shih and Elkobi. Fourth, it asks whether AI is the vindication of financial capitalism, citing Epoch AI data on cost declines and Karthik Tadepalli's estimate that cost fall as a function of spending does not stand out, then quoting James Gilliland's argument that a boom in industrial capacity from robotics would look like vindication of neoliberalism. Fifth, it explains Japan's land readjustment approach to housing and infrastructure, drawing on a World Bank case study by Matsui Minoru, in which existing tenants get a new home on the same spot at the same rent, and describing how no one is expropriated. Sixth, it notes phones are bad, citing Pew data on social media harming democracy and Greece's recent ban on social media for children under fifteen. Seventh, it argues China is losing market share in low-end manufacturing despite the leadership's wishes, with Bangladesh, Vietnam, and Mexico gaining at China's expense. Eighth, it argues rent control is bad, citing John Burn-Murdoch's synthesis of one hundred twelve studies showing rent control helps existing tenants in a narrow way but removes incentives for developers to build and landlords to upkeep, driving rents up elsewhere and some people to homelessness.

Love with nowhere to go from Aeon is an essay by Derek van Zoonen arguing that grief is not only for the dead, and that when we are unable to honour other profound losses, we become strangers to ourselves.

Eleven outstanding film and television reviews by Tyler Cowen and Alex Tabarrok from Marginal REVOLUTION is an archive post, with selections written by ChatGPT. The list includes five reviews by Tyler Cowen and six by Alex Tabarrok. Cowen's selections include The Devil Wears Prada: A Straussian Interpretation, Toni Erdmann, Misunderstood Masterpiece, Her, Technology in Kubrick's 2001: A Space Odyssey, and Borgen Season Four. Tabarrok's selections include Fleabag and Killing Eve, Perfect Days, Contagion, Daredevil and the Doctrine of Double Effect, The Remarkable Life of Ibelin, and The Gaslighting of Parasite. The post emphasizes the quality of the criticism: original interpretations, close attention to the work, and arguments worth considering even when one disagrees, and notes that several original posts reveal major plot developments while the descriptions keep those revelations to a minimum.

Book Review: You Could Do Something Amazing with Your Life [You Are Raoul Moat] - Andrew Hankinson ★★★★☆ from Terence Eden's Blog gives the book four stars out of five. The review says it is hard to write a book in the second person and even harder to make it this good. Andrew Hankinson attempts to use Raoul Moat's own words to build up a picture of the man, written with sympathy but without being sympathetic. The stream of consciousness is interrupted only occasionally by the voice of god, which probably isn't god, to provide clarity to an unreliable narrator. The paranoia and delusion come streaming off the page in a tone that exudes reasonableness. Almost every sentence is a challenge to the reader about whether this man is telling the truth and what society can do when someone's model of the world is so broken. The review notes the constant drumbeat of Moat's claim he was repeatedly victimised by police, and that there does appear to be a conspiracy of silence around how he died. It asks whether missing a few meetings with a social worker should result in being detained against one's will, and whether readers are happy to accept his attacks as a price worth paying for freedom. The review calls it excellent and provocative.

Marissa Mayer Can't Keep Pretending 'Gender Doesn't Matter' from 404 Media opens with the widely discussed New York Times interview in which Marissa Mayer, former president and CEO of Yahoo, asked her own AI assistant, the app Dazzle, to name three things she loves about her ten-year-old daughter, and the assistant answered that her daughter is empathetic, creative, and an animal lover. The article says what Mayer has to say about gender in tech is more out of touch than that. Mayer told Steven Levy in a 2015 interview that she never plays the gender card, and repeats similar sentiments in Sunday's interview, saying passion is a gender neutralizing force. The article notes that male AI figureheads are posting slop videos on main, half of women in tech leave the industry by thirty-five, women earn eighty-four cents to the dollar, and women founders receive half the funding of male founders. It reports that the Grace Hopper Celebration, one of the biggest gatherings of women and non-binary technologists, was canceled earlier this month for lack of registrations and sponsorship, after a 2023 conference where hundreds of men flooded the event and attendees reported being shoved, harassed, and cut in front of. Women Who Code shut down in 2024 after eleven years. The article then turns to the Trump administration's executive order ending DEI programs, the NIH termination of nearly two thousand three hundred federal research grants that disproportionately hurt women, and DOGE's cancellation of funding for research on pregnancy, reproductive health, chronic illness, and vaccine safety for women. It concludes that in a way Mayer is right that drinking one's own Kool-Aid is a non-gendered issue, and notes Sam Altman's earlier AI podcast suggestion drew similar ridicule.

Game Boy Color on My Conference Badge from Simon Willison's Weblog reports that the GitHub Universe conference badge is a full-blown color screen computer. The author hacked around with it and ran into Steve Francia, who worked with Gemini to get a full NES emulator working, then went a step further and deployed an emulator for both the Game Boy and the Game Boy Color. If one can obtain the ROM, one can run The Legend of Zelda: Link's Awakening on the conference badge. A neat feature is that the badge can be turned upside down to use game controller touch buttons built into the body.

Tuesday assorted links from Marginal REVOLUTION contains six links: which musical styles smart people prefer, University of Austin job ads, Walter Koenig's death at the New York Times, one take on a new computational result along with one overstated Chinese interpretation and another extreme and unwarranted view, a new research method described as Arjo Klamer's revenge, and a piece asking how the Nobel Prize became high status.

The window spans corporate accountability around microplastics, a pivotal week in AI safety and capital, a broad sweep of breaches and enforcement actions, political transitions in Israel, Lebanon, and Serbia, and a cluster of Nautilus science features on parks, bats, spiders, memory, and food.

  1. Industry Executives Knew For Decades That Microplastics Were Infiltrating the Human Body
  2. Roundup #90: In which Noah joins a Japanese VC firm
  3. NYT debate on the rules for AI companies
  4. The half-life of disappointment
  5. Love with nowhere to go
  6. Roblox downloaded 145GB models from us (including adult content) without asking
  7. Anthropic S-1, The Cost of Compute, Who Is Paying?
  8. Why the Majority of Israel’s Youngest Voters Identify as Right Wing
  9. Denmark Gives Up On Replacing Microsoft Office With Collabora
  10. Nvidia backs Indian startup helping companies run AI models with fewer GPUs
  11. SDK 0.31: a Ruby gem, Flutter parity, live BLE signal and sessions that heal
  12. Hiding Capabilities in AI Models
  13. Eleven outstanding film and television reviews by Tyler Cowen and Alex Tabarrok
  14. Software may be getting worse, but it used to be horrible
  15. Can Amal Help Lebanon Find a Way Out of War?
  16. Take a Look at the Country’s Newest National Park
  17. SDK 0.31: a browser runs a full Offline Protocol node
  18. Book Review: You Could Do Something Amazing with Your Life [You Are Raoul Moat] - Andrew Hankinson ★★★★☆
  19. Anthropic Versus Anthropic
  20. Opera introduces a new web standard to let users deploy AI locally in their browser
  21. General Compute raises $100 million Series A, countering the neocloud disaster narrative
  22. Anduril, Qualcomm launch BlueChip, a new kind of US-made laptop
  23. Introducing the Cloudflare PHP Capture the Flag challenge
  24. What took the Internet offline in Q3 2026
  25. Billionaire Palmer Luckey isn’t leaving California over the wealth tax
  26. Apple introduces the new 14-inch MacBook Pro with M6
  27. Marissa Mayer Can’t Keep Pretending 'Gender Doesn't Matter'
  28. Australia’s spies are sharing intelligence with hacking targets — and want them to share back
  29. CISA warns of PHP bug exploitation attempts, gives federal agencies days to patch
  30. Moscow faces internet disruptions as foreign leaders gather for summits
  31. Japan internet giant GMO says hackers posted stolen customer data
  32. Malicious apps prey on Russians’ anxiety about fuel prices, drone strikes
  33. It's not the bots, Flock
  34. Apple rolls out iOS 27.2 with new Siri AI languages and security fixes
  35. When Texas Bat “Cities” Look Like Thunderstorms
  36. Your Decisions Are Probably Not Your Own
  37. The US Government Is Hiding Secret Court Ruling About How It Spies on Americans
  38. Serbia’s Victorious New Democrats Face an Old Question
  39. Apple unveils new 14-inch MacBook Pro with M6 chip
  40. US government says hackers are now exploiting critical PHP web bugs
  41. Ford Fathom will be the cheapest American EV capable of powering a house
  42. University of New Mexico hospital diverts ambulances after cyberattack
  43. Meta AI director Rohit Patel leaves to take on Microsoft Office and Google Workspace
  44. In China, the AI Bubble Has Already Burst
  45. Hackers repeatedly breached suppliers holding UK police data, watchdog finds
  46. Apple Updates MacBook Pro With M6 Chips
  47. How Jumping Spiders Sleep
  48. The long road to bitcoin payments...
  49. Genie 4
  50. Google’s Genie 4 world model adds audio, more realistic worlds, and real-time steering
  51. Startups in a World of Abundant Space Access
  52. Tuesday assorted links
  53. A 28-Year-Old DOGE Operative Gutted USAID. HR Had No Record of Him Working There
  54. Police Are Using AI to Pull over Drivers for 'Suspicious' Behavior
  55. Ghostty 1.4.0
  56. Cybersecurity CEO Indicted For Allegedly Scamming Millions From Ransomware Victims
  57. Frontier AI labs say their models refuted the Exponential Time Hypothesis. What does that mean?
  58. How to protect your privacy while traveling
  59. Apple's Foldable iPhone Duo Draws Early Complaints of the Screen Cracking
  60. The Teenage Girl Who Remembers Everything
  61. Robots Made In America
  62. Disruptive raises $10B fund to make big bets on at least 10 companies
  63. Anduril Partners With Qualcomm to Build Laptops On US Soil
  64. US indicts Malaysian national after Madagascar scam operation uncovered
  65. Game Boy Color on My Conference Badge
  66. Google's Genie 4 World Model Adds Audio, More Realistic Worlds
  67. There’s a New, International Food Pyramid in Town
  68. Re:Build, which spun out Slate Auto, raises $2.1B
  69. Nefilim, Hive ransomware operator given five-year-plus sentence after guilty plea