↵ select ↓ ↑ navigate esc close

Daily Briefing: November 8, 2026, evening

Daily Briefing ·

Listen: https://serve.buildtall.systems/4ebeb30eb18da7fc1420326df039b22ae2cca1e8defef1c00629666a30234d16.mp3


The briefing for November 8, 2026 covers twenty-two articles.

The first group concerns AI systems and their consequences.

Claude Code's 'Don't Ask Again' File Exposes Passwords and Keys in 1,122 Public Copies on GitHub reports on a file created by Anthropic's AI coding assistant. When a user approves a command with the "Yes, and don't ask again" option, Claude Code can save that command word for word, including any password or API key, in a project settings file called .claude/settings.local.json. Software security company Apiiro found 1,122 public copies of that file on GitHub containing at least one secret, and 1,379 different secrets altogether. The secrets include 379 database passwords, 43 SSH logins, 42 tokens for GitHub, GitLab and npm, and 23 AWS keys, along with an admin-level cloud key from a database vendor and the owner password of a medical-records database. The article warns that deleting the file does not undo the leak because older versions remain in git history and in forks. Researchers advise rotating every password, token, and key before cleaning up. Claude Code does try to keep the file out of git by adding it to a user's global git excludes the first time it is written, but that protection stays on one machine and does not travel with a project. In 539 of 1,105 traced repositories, a commit listing Claude as a co-author changed the file, suggesting Claude itself often commits the file. Anthropic's head of Claude Code, Boris Cherny, responded that local settings are not meant to be checked in, that permission rules are old and auto mode is safer, and that users should enable secret scanning. He also said Anthropic is looking into basic secret scanning and making auto mode catch the problem. The article notes that a bug report filed in August 2025 was closed as not planned in January after sixty days without activity, and three further reports received no reply.

Parents Sue Google Over AI Safety Risks On School Chromebooks describes a lawsuit filed in California Superior Court on Monday. The New York Times reports that a thirteen-year-old student referred to as Emma made plans to end her life after months of chatting with Google's artificial intelligence on a school-issued Chromebook. Her parents first learned of the danger in mid-March when the school warned that she was in imminent danger of self-harm. Her mother found a suicide note on Google Docs, and the family took Emma for an emergency psychiatric evaluation; she now attends therapy weekly. The complaint says Emma, a seventh grader in Charlotte, North Carolina, had become convinced she had multiple mental health disorders after interactions with AI Mode, a chatbotlike feature on Google Search, and had searched for what a suicide note is and told the AI she was planning her own funeral. The suit argues that Google hastily integrated AI features into apps used on K-12 Chromebooks without notifying parents of risks, and that some students had AI access enabled by default. While school administrators can toggle apps like YouTube and Gemini, they cannot disable AI Mode and AI Overview embedded in Google Search. This week Google announced a policy of disabling AI tools by default for schools, and schools that re-enable search may choose between existing embedded AI features and a new educational AI feature, but Google still offers no option to turn off AI altogether in search because search and AI are deeply integrated. The case is part of a wave of litigation over Big Tech's effects on young people.

OpenAI VP: How AI Changed My Family’s Fight Against Cancer | Ashley Alexander is a podcast episode from Behind the Craft featuring Ashley, VP of Product at OpenAI for health. The host and Ashley discuss how both use AI to help their parents fight cancer, including preparing for doctor visits, researching treatments, and coping with emotions as caregivers. Ashley's parents were both diagnosed with cancer within a single week, and she took a month off from work without regret. The episode's takeaways include giving ChatGPT a patient's full medical history to support informed discussions with doctors, using ChatGPT to flag possible conflicts before treatment, keeping ChatGPT open during telehealth appointments to look up unfamiliar terms, and storing a parent's records in a separate project with consent and redaction. Ashley describes keeping a one-page care summary and setting up a recurring ChatGPT task to check ClinicalTrials.gov weekly for new research. The episode also emphasizes making time for mental health and sharing the situation, which helped Ashley find a Stanford specialist through a friend's father-in-law. The article mentions Granola, a tool used to record medical conversations with the care team's permission.

The next group covers two articles on Nvidia CEO Jensen Huang.

Nvidia says it’s investigating CEO Jensen Huang’s relationship with possible Chinese spy reports that Huang admitted in a memo to the company that he had an affair.

Nvidia CEO Admits Affair With Alleged Chinese Spy. Investigation Opened adds details from CNBC and Axios. Huang told Nvidia staff in an email on Sunday that he had an affair and that the board had opened an independent investigation into the spying allegations. He wrote that he was unfaithful to his wife, that his actions hurt her and their children, and that there is no excuse. The board said an independent committee assisted by outside counsel will investigate, that it has no reason to believe national security or company secrets were compromised, and that it expressed confidence in Huang's leadership. The article notes that Huang has been a key voice in Washington pushing to loosen restrictions on AI chip sales to China, arguing that keeping American technology in the Chinese market strengthens U.S. competitiveness and prevents China from developing more advanced chips. An Nvidia spokesperson said Huang was committed to continuing as CEO without interruption. U.S. lawmakers have requested a closed-door briefing, but Huang has not confirmed he will appear, and a company spokesman declined to comment. The spokesman also said Huang's interactions with the alleged Chinese spy were consistent with meetings he has with Nvidia customers.

The third group concerns electricity grids and data centers.

Coordinated disturbances haunt grids worldwide. Here’s how researchers solved the mystery. states that coordinated power drops have been haunting power grids worldwide and that a team of researchers in Germany found the culprit.

Do Data Centers Have a Smaller Effect on Electricity Prices Than We Think? reports on a study by researchers at Harvard University and the University of Pennsylvania. The researchers calculated that data centers were responsible for just four percent of utility rate increases over the last five years. The article acknowledges that some regions covered by grid operator PJM, including Pennsylvania, Virginia and Ohio, have seen data centers push costs onto residential customers, but multiple studies have failed to find a strong connection between data centers and overall U.S. electricity prices. An MIT analysis found data centers increased average residential electricity prices by 2.1 percent between 2010 and 2024. Research from the Brattle Group and Lawrence Berkeley National Laboratory found data centers were decreasing costs in certain regions, and an Electric Power Research Institute study found a doubling of data center capacity between 2019 and 2024 resulted in a four percent drop in home electricity prices. The Harvard and Penn study found a higher impact on industrial electricity prices, with data centers responsible for fifteen percent of the price increase between 2020 and 2025. Wharton professor Serguei Netessine, one of the study's authors, says so far there is no evidence data centers increase prices all that much, and that prices are rising due to gas prices, an aging grid, and recovery from wildfires and hurricanes. If data center growth continues, residential prices could be affected more later. The article also notes public opposition that goes beyond price, including noise, pollution, disturbed neighborhoods and landscapes, and concerns about AI and job loss; a Pew Research Center survey found more than half of Americans say data centers are mostly bad for the environment.

The next group covers free and open-source software legacy and security.

Aaron Swartz Would've Been 40 Today. His 'Legacy Lives On' Says FSF reports on Aaron Swartz Day. FSF deputy director Sean O'Brien writes that Swartz would have turned forty this year. The Internet Archive hosted a hackathon Sunday in San Francisco celebrating projects he worked on, including SecureDrop, OpenLibrary and Creative Commons, and projects he inspired such as OnionPress and the EFF's Atlas of Surveillance documenting police tech use. Free Law Project supported the event, and Portland had a screening of The Internet's Own Boy. The article details Swartz's co-founding of Creative Commons, where he worked with Lisa Rein on RSS-enhanced HTML behind Creative Commons licenses, and the Quantifying the Commons project. He created Tor2web with Virgil Griffith, and Onion Press is a Tor-based publishing system for decentralized censorship-resistant publishing. Swartz collaborated with John Gruber on Markdown in 2004, which is now widely used in documentation and community websites, and OpenAI recommends Markdown for organizing instructions to ChatGPT models. He also developed web.py and contributed to RSS 1.0, developing web.py while rewriting Reddit software after Reddit acquired his company Infogami. The article says Reddit recently licensed its content to AI companies, and the FSF deputy director contrasts the alleged piracy by giant technology corporations with Swartz's criminal prosecution for downloading academic research he wanted to distribute without monetary gain. Federal prosecutors charged Swartz in 2011 after he downloaded millions of academic articles from JSTOR through MIT's network, and he died by suicide in January 2013 at twenty-six. The article argues that companies benefiting from his legacy should respect his access-to-knowledge principles by giving users freedom and control.

Norwegians Release FOSS GNOME Software for Music Recording, Internet Radio, and Cameras covers free Norwegian open-source projects. Slashdot reader ole describes GarageJam, a GNOME music recording studio for composing, recording and sharing music, with version 8.0.1 available in C and Python. It records audio immediately to Ogg Vorbis files, edits metadata in .gjxml XML files from a GTK/GNOME wizard, and is built on the Gingerblue Recording Studio API. There is also GarageJam Studio, a commercial music and audio recording app from Aamot Innovation in Norway. GNOME Media puts internet radio and photographs on one OpenStreetMap, built from GNOME Radio 85.0.1; users can click a station to listen, see counts and menus for cities with several stations, open folders of pictures or import from a camera, view photos with GPS EXIF data on the map, double-click thumbnails, and filter stations and map together. GPhoto 3.0.1 is built on libgphoto2 2.5.34, which now supports more than 2,500 camera models. Skolelinux 13.6.0 is a complete free-software operating system for schools built on Debian 13, including office suite, web browser, educational programs and network services out of the box.

Python's CVEs Triple. Anthropic Funds a New Security Developer reports that CVEs for Python and its package manager Pip have increased threefold in the last year, and malware published and removed from PyPI is up sixty percent, according to Python security developer in residence Seth Larson. Securing Python is more complex because of the evolving role of AI in security risks and response. The Python Software Foundation has brought on Security Developer Brett Smith to address vulnerability remediation, provide vulnerability data as a CVE Numbering Authority, remove malware, scale PyPI's autonomous quarantine program, and streamline processes to handle the new vulnerability report volume. The role is funded by an ecosystem security grant from Anthropic. Curl's Daniel Stenberg reported in June that Curl's CVEs have also tripled.

A set of shorter technology industry items follows.

Ninja’s AutoBarista is impressively versatile, but one feature was a dealbreaker for me reports that the AutoBarista's Grind iQ technology allows the machine to analyze coffee beans to inform how it pulls a shot of espresso.

TechCrunch Mobility: The Cybertruck experiment is over is a newsletter welcoming readers to coverage of the future of transportation and the role AI is playing in it.

He arrested a hacker, then spent 18 months clearing his own name. Now he’s a cybersecurity VC. states that Stephen Ward, a former Secret Service agent and Home Depot CISO, is preparing a second fund of roughly two hundred fifty million dollars for Brightmind Partners.

SNL has a new impression of Elon Musk, ‘the worst thing South Africa has ever done’ says that Saturday Night Live seems to have its knives out for tech executives.

Sequoia partners reportedly blindsided when Roelof Botha joined SpaceX board reports that a new report claims Sequoia's recent changes undo the legacy of the firm's former steward.

If Snap’s Specs are still ridiculously expensive, they’re also a lot more comfortable than they used to be reports that Snap's Specs, which have been in development for many years, are going to market this fall, and asks whether they will be a hit with users.

The remaining articles stand outside those groupings.

Ambiguity or clarity? from Seth's Blog argues that choosing how much ambiguity to handle can define a career and shape social and cultural interactions. The desire for clarity is based on safety and risk, because knowing what to expect avoids responsibility for things outside one's control, and clarity pushes things into predictable categories. Ambiguity, by contrast, is the nature of the future, since interactions with people, complex systems and what has not happened yet are unpredictable, and that range of outcomes is where possibility lies. Phrases like I'm not sure, This might not work and We'll be okay are cited as examples, while pretending the ambiguous is certain is short-term bravado that works in movies and politics but not real life. The article closes with the chance to embrace the soft edges of perhaps on the way to making things better.

Some thoughts on restaurant menu QR codes from Terence Eden’s Blog opens by comparing complaints about QR restaurant menus to complaints about hotels replacing proper keys with plastic keycards. The author and his wife took a seven-week Interrail trip through Europe and found electronic menus available in many restaurants; they were brilliant. An electronic menu allows instant translation by selecting text, often already has translations, legally mandated allergen information, and stock availability, and photos or videos of dishes are useful. Online ordering avoids the waiter mishearing a request and supports notes to the kitchen about allergies, and payment can be made through the web with bill splitting. The author dismisses objections one by one: every restaurant had fast free Wi-Fi, no required one-time app, upselling is comparable to a waiter recommending expensive wine, and paper menus were still available for those who asked. In the worst case, the advice is simply to walk out, just as one would leave a restaurant for a rude waiter or a rat.

Reverse engineering a vintage diode-transistor logic chip from Ken Shirriff's blog describes identifying an unknown integrated circuit from CuriousMarc's collection. After opening the chip and studying it under a microscope, the author determined it was the ST680A, a member of the Signetics 600 series from 1965, implementing four NAND gates with diode-transistor logic. The article explains the history of the plastic DIP package, which Signetics introduced in late 1965, and the process of decapping the epoxy package, cleaning the die with sulfuric acid, and reconstructing the missing metal layer to infer the circuit. The chip contains NPN transistors, diodes and resistors; the NAND gate works by using input diodes and two transistors so the output is low only when both inputs are high. Signetics used the same silicon layout for at least three chips, the 680A quad NAND gate, the 670A triple three-input NAND gate and the 616A dual five-input NAND gate, changing only the metal layer. The article also traces Signetics history from a Fairchild spinoff founded in 1961, its 1962 first commercial DTL family, investment from Corning Glass, competition from Fairchild, the 555 timer in 1970, acquisition by Philips in 1975, and eventual spin-off as NXP.

Crisis and Resilience in Historical and Modern Societies from Cliodynamica reports that Peter Turchin's research group at the Complexity Science Hub ran a workshop on Thursday and Friday. The workshop aimed to foster collaborations between researchers studying societal crisis and resilience in historical and contemporary case studies. The group is nearing the end of the first year of a three-year program funded by the Austrian Science Fund, studying dynamics of resilience and instability using historical and contemporary data with computational models, aiming to determine whether human societies are dynamical systems that can be studied and forecast with complexity science methods.

Sunday assorted links from Marginal REVOLUTION collects nine links. The listed items include Among the Wagner fanatics, Does AI instead commodify jobs?, The digital world is making Islam more diverse, An affair with an economist exposed the calculus of modern dating, How Much Economic Value Do AI Systems Produce per Unit of Energy?, John Gray will be writing a column for Unherd, How France took over basketball, If existing AI capabilities are widely adopted, they could double research output in some fields, and Canvas is why students prefer AI over you.

Salon, One of the Web's Pioneering News Outlets, Shuts Down After 31 Years reports that Salon.com will stop publishing after thirty-one years. The announcement on the site under the byline Salon remembers that the publication was an idea about what journalism could be, born when the internet was still largely unexplored. Launched in November 1995, Salon pioneered a form of digital journalism that brought together politics, culture, criticism and ideas, and its writers and editors helped shape conversations about forces changing lives. The announcement says the business realities of independent digital journalism have made the publishing model unsustainable. A spokesperson said nineteen full-time employees and a handful of contractors were affected. Slashdot reader oumuamua remembers Salon as one of the internet's first digital-native publications, and the site linked to Salon many times, including a 1998 story comparing Microsoft and the Catholic Church and Linus and Martin Luther. Longtime staff writer Amanda Marcotte wrote that she and everyone are the latest victim of AI's predations, and Variety notes Salon joins Buzzfeed and Gawker in failing to monetize cultural and topical reporting for a general audience.

Across the window, the articles return repeatedly to the pressure AI systems place on security, children, healthcare and public infrastructure, while also recording the closure of a pioneering digital publication and the legacy of figures who built open tools.

  1. Ambiguity or clarity?
  2. Person of Interest
  3. Aaron Swartz Would've Been 40 Today. His 'Legacy Lives On' Says FSF
  4. Some thoughts on restaurant menu QR codes
  5. Introducing PREDYX HUB
  6. Prediction Markets INSIDE Personal Finance Apps (FT, Stephanie Stacey)
  7. All LNbits users should stop their LNbits servers IMMEDIATELY
  8. OpenAI raises $11B at a $1.25T valuation
  9. Ninja’s AutoBarista is impressively versatile, but one feature was a dealbreaker for me
  10. Is AI Now Good Enough To Cure Cancer? | YC Paper Club
  11. OpenAI VP: How AI Changed My Family’s Fight Against Cancer | Ashley Alexander
  12. Python's CVEs Triple. Anthropic Funds a New Security Developer
  13. Tesla Full Self Driving Review
  14. Tesla Hasn't Had a Successful New Product Launch Since JB Straubel Stepped Down
  15. Reverse engineering a vintage diode-transistor logic chip
  16. Norwegians Release FOSS GNOME Software for Music Recording, Internet Radio, and Cameras
  17. TechCrunch Mobility: The Cybertruck experiment is over
  18. Crisis and Resilience in Historical and Modern Societies
  19. Sunday assorted links
  20. He arrested a hacker, then spent 18 months clearing his own name. Now he’s a cybersecurity VC.
  21. Do Data Centers Have a Smaller Effect on Electricity Prices Than We Think?
  22. SNL has a new impression of Elon Musk, ‘the worst thing South Africa has ever done’
  23. Goldman Sachs Is Pulling Claude Code from Its Bankers and Traders
  24. Claude Code's 'Don't Ask Again' File Exposes Passwords and Keys in 1,122 Public Copies on GitHub
  25. Salon, One of the Web's Pioneering News Outlets, Shuts Down After 31 Years
  26. Sequoia partners reportedly blindsided when Roelof Botha joined SpaceX board
  27. Nvidia Board of Directors Is Investigating Jensen Huang
  28. Coordinated disturbances haunt grids worldwide. Here’s how researchers solved the mystery.
  29. Parents Sue Google Over AI Safety Risks On School Chromebooks
  30. If Snap’s Specs are still ridiculously expensive, they’re also a lot more comfortable than they used to be
  31. Nvidia says it’s investigating CEO Jensen Huang’s relationship with possible Chinese spy
  32. Nvidia CEO Admits Affair With Alleged Chinese Spy. Investigation Opened