News

News from November 6, 2026

In my view, Tesla’s shift, Apple’s filters and the demands on TikTok show how business choices shape the technology we use.

The day in three lines

  1. Elon Musk said Tesla will wind down Cybertruck production in 2027 to make more Cybercabs.
  2. The European Union demanded changes to TikTok recommendations, saying repeated content may harm users’ well-being.
  3. A researcher found about 100,000 terms missing from iPhone keyboard suggestions; Apple says it aims to prevent offensive recommendations.
  1. Tesla to wind down Cybertruck production

    Elon Musk said Tesla will wind down Cybertruck production in Texas around mid-2027 to make more Cybercabs.

    Hacker News863 points · 1,727 comments

    Quote
    Musk said in an X post the production wind-down would enable the automaker to accommodate increased output of its newly launched Cybercab model,
  2. A Strange Quantum Theorem Has Killed Einstein’s Reality

    Researchers trying to defend Einstein’s everyday notion of reality keep finding holes in his argument.

    ScienceQuanta Magazine

    Quote
    Albert Einstein fought to save our everyday notion of reality from the strangeness of quantum mechanics. But those who set out to aid him keep poking holes in his argument.
  3. The Download: a humanoid cleans house and Samsung builds for the AI boom

    A humanoid robot cleaned the author’s apartment, and she described the experience as a taste of the future.

    TechMIT Technology Review

  4. Partnering with Hacktron to find vulnerabilities in OpenAI’s systems

    OpenAI is working with Hacktron to find vulnerabilities, test connections to outside services, and strengthen defenses using advanced artificial intelligence.

    AI labOpenAI

    Quote
    OpenAI is partnering with Hacktron to find vulnerabilities, test connections with third-party services, and strengthen defenses with frontier AI.
  5. O boletim Notícias FAPESP e a origem da rede Scielo

    The SciELO network began in 1996 to make Brazilian research easier to find and share, offering scientific journals free online.

    BrazilPesquisa FAPESP

    Quote
    Há 30 anos, num almoço em Caxambu (MG), nascia o Scielo para melhorar a difusão da ciência brasileira
  6. VFlash: Fast, Accurate Visual Planning via Flash Primitive Decoding

    A study presents a method that predicts visual plans all at once, reducing errors and speeding up robot motion forecasts.

    Papernot peer reviewed · 95 votes

    Quote
    Models for physical AI tasks, such as robotics and autonomous driving, need not reason solely in language. They can use visual plans--points, bounding boxes, and trajectories--to represent spatial goals and anticipate motion.
  7. Record-breaking, fast-spinning magnet could help uncover new physics

    Researchers made a tiny levitating magnet spin extremely fast and steadily, which could help measure magnetic fields and investigate dark matter.

    ScienceNew Scientist

    Quote
    A magnet barely wider than a human hair is one of the fastest spinning objects ever and could be useful for both fundamental physics experiments and in quantum technology
  8. The Cybertruck’s Death Is Actually Great News for EVs

    Tesla canceled its massive electric vehicle as smaller models spread, a shift that could benefit the US battery supply chain.

    TechWIRED

    Quote
    Tesla’s massive EV is the second within the last year to get the ax, while smaller vehicles are proliferating. It could be a huge benefit for the US battery supply chain.
  9. ToyBlocks: A controlled world for robot learning and evaluation

    Ai2 created a robot-arm-and-block simulation that generates training examples and lets researchers test how changes in data affect performance.

    AI labHugging Face

  10. Polvos adotam duas estratégias para passar por buracos minúsculos

    Scientists found that octopuses change the order in which they arrange their arms and bodies as openings get narrower.

    BrazilFolha, Ciência

    Quote
    Os polvos são famosos por passar por frestas aparentemente impossíveis.
  11. Blender's UI and Runtime: Much Faster Than GPUI

    The project offers a graphics-card-accelerated interface separate from Blender and claims it outperforms a competing interface.

    Hacker News686 points · 278 comments

    Quote
    Blender's GPU Accelerated UI, Decoupled from it's original codebase. Supports Rust, Python, JS, React, C & C++. Twice as good as GPUI - ppmpreetham/BUI
  12. We're finally learning how your cholesterol levels shapes your dementia risk

    The text says cholesterol protects neurons and researchers are investigating its link to dementia risk.

    ScienceBBC Science Focus

    Quote
    A quarter of our body’s supply can be found in our skulls, protecting our neurons
  13. How Do You Land on a Moon With Almost No Gravity?

    The MMX mission will explore Phobos, Mars’s tiny moon, and attempt to land, collect samples, and bring them back to Earth.

    TechIEEE Spectrum

    Quote
    Last August, looking up from the edge of an ancient river delta on Mars, the Perseverance rover ’s electronic eye spied an oblong shadow crossing the distant, glowing Sun.
  14. LiquidAI: LFM3-1.7B

    A compact artificial intelligence model handles long texts and was trained to use tools and organize responses.

    AI labOpenRouter

    Quote
    A compact 1.7B-parameter text model with a 256K-token context, post-trained for agents, tool calling, and structured outputs. It reasons before it answers. Prompts and outputs may be retained and used...
  15. Apple restringe sugestões de termos como 'Trump ditador' e 'Joe sonolento' no iPhone

    A researcher found about 100,000 terms missing from iPhone keyboard suggestions, and Apple says it aims to prevent offensive recommendations.

    BrazilG1, Tecnologia

    Quote
    Pesquisador identificou cerca de 100 mil termos que não aparecem nas sugestões do teclado do iPhone, incluindo expressões políticas em diferentes idiomas; Apple diz que busca evitar recomendações ofensivas.
  16. SLA: Scaling Spatial-Language-Action Model via Continual Pre-training

    A robot model learns three-dimensional positions and movements from varied experience while keeping its operation unchanged when performing tasks.

    Papernot peer reviewed · 91 votes

    Quote
    Reliable robotic manipulation requires both semantic understanding and precise control in three-dimensional space.
  17. Hera spacecraft photographs asteroid from 37 miles away

    The spacecraft is traveling to study the effects of the 2022 collision with the asteroid Diphamos.

    SciencePopular Science

    Quote
    The ESA’s spacecraft is en route to study the aftermath of DART's collision with Diphamos in 2022.
  18. Doctors pull knife from man's eye that went through his brain—and he's fine now

    Doctors removed a knife that pierced a man's eye and brain; he recovered, but surgeons had to remove the eye.

    TechArs Technica

    Quote
    It's the exact path, not the apparent severity that matters most, doctors say.
  19. iPhone tem lista secreta com mais de 100 mil palavras que evita sugerir

    An internal iPhone list prevents the keyboard from automatically suggesting over 100,000 words and phrases, including swear words and political statements.

    BrazilTecnoblog

    Quote
    Lista interna do iOS impede que termos sejam sugeridos pelo corretor do teclado em diferentes idiomas.
  20. Dutch 'idealist' bank ASN blocks non-Google Android, including GrapheneOS

    A user reports that the ASN Bank app works on a Pixel 6 Pro without Google Play or Google Play services.

    Hacker News622 points · 406 comments

    Quote
    Is there an existing issue for this? I have searched the existing issues App name ASN Bank Link to app https://play.google.com/store/apps/details?id=nl.devolksbank.asn.bankieren App version 2.40.2...
  21. This may be the best evidence yet that dinosaur blood can fossilize

    A study says it found traces of blood cells in 115-million-year-old dinosaur bones, though some researchers are not convinced.

    ScienceScience News

    Quote
    Imaging and chemical analysis of a 115-million-year-old dinosaur fossil points to blood cells. But some aren't convinced.
  22. OpenAI valued at $1.25 trillion, as rival Anthropic nears its IPO

    According to Bloomberg, OpenAI was valued at $1.25 trillion after raising another $11 billion.

    TechThe Verge

    Quote
    OpenAI is now valued at $1.25 trillion in a new round of funding that has raised another $11 billion, according to Bloomberg.
  23. Blue Origin adia tentativa de lançamento do foguete New Glenn para 2027

    Blue Origin told employees the rocket’s next flight is expected in January 2027, not this year.

    BrazilFolha, Ciência

    Quote
    A Blue Origin não tentará mais lançar o foguete New Glenn neste ano. Na última quarta-feira (4), funcionários foram informados de que a previsão do próximo voo é janeiro de 2027.
  24. Harness-S: Learning from Sibling Harnesses for Generalizable Agent Training

    A study trains agents with auxiliary versions of an execution system and reports that they improve even when returned to the original system.

    Papernot peer reviewed · 55 votes

    Quote
    Harness evolution from a common base exhibits Locality, Complementarity, and Nonstationarity: variants repair different failures, and their utility changes as the policy learns.
  25. This brain change can make foxes act like pet dogs

    A study found fewer aggression-linked brain cells in tame foxes selectively bred in Russia over generations.

    ScienceScientific American

    Quote
    Using data from a 1950s Russian experiment, researchers are getting “to the heart of brain programming” in foxes
  26. Crusoe sells troubled Wyoming data center project to Google

    Crusoe sold Google an unfinished computing facility in Wyoming, and Google plans to invest $17 billion to develop the site.

    TechTechCrunch

    Quote
    The deal highlights the challenges of developing massive AI data centers.
  27. Japão anuncia pacote emergencial de cibersegurança após onda de ataques

    Japan will invest $63 million in artificial intelligence and specialist teams to find network weaknesses, help affected companies, and prevent cyberattacks.

    BrazilG1, Tecnologia

    Quote
    Plano prevê investimento de US$ 63 milhões (R$ 345,24 milhões), uso de inteligência artificial para identificar vulnerabilidades em redes e criação de equipes especializadas para apoiar empresas afetadas e reforçar a prevenção de ameaças…
  28. Britain to make salary disclosure in job ads mandatory

    No summary: the source’s text could not be read.

    Hacker News565 points · 430 comments

  29. Fabric molds could bring material-saving concrete beams within reach

    An MIT study made a beam with a sewn fabric mold, using one-third the concrete and half the steel of conventional beams.

    ScienceMIT News

    Quote
    MIT engineers cast a concrete beam using a sewn fabric mold with one-third the amount of concrete and half as much steel as a conventional beam.
  30. Agents reveal secrets when offered to team up with each other

    In tests, AI coding agents sometimes revealed private credentials and ran unknown code to get help from other agents.

    TechThe Register

    Quote
    If you wouldn't put it on GitHub, don't give models access to it
21 more
  1. AMD agora prioriza data centers; mas e os jogos e os PCs?

    Data centers now bring in most of AMD’s revenue, but the company has not indicated it will leave gaming and PCs behind.

    BrazilTecnoblog

    Quote
    Aplicações de IA aumentaram demanda por data centers que, por sua vez, agora são principal fonte de receita para a AMD.
  2. Random Is Almost Enough? Stable and Low-Cost Benchmark Compression via BestRand

    A study introduces BestRand, a method that selects a small part of a test set to evaluate models at lower cost and with less risk of large errors.

    Papernot peer reviewed · 50 votes

    Quote
    Benchmark compression reduces the often prohibitive time and cost of model evaluation by evaluating models on a small benchmark subset that closely matches full-benchmark results.
  3. M30’s Sparkling Stars

    A Hubble image shows the star cluster Messier 30, cataloged by Charles Messier among objects that could be mistaken for comets.

    ScienceNASA

    Quote
    This NASA/ESA Hubble Space Telescope image released Nov. 5, 2026, provides a lesson in galactic history.
  4. I watched a humanoid robot clean my apartment. It felt like a taste of the future.

    The author signed up for a new cleaning service using humanoid robots in San Francisco.

    TechMIT Technology Review

    Quote
    I’m not the sort of person who typically pays for a house cleaner. I’m too frugal, and I kind of enjoy putting on a podcast and going to town on my tub with a Magic Eraser.
  5. UE exige que TikTok mude feeds personalizados por possíveis efeitos nocivos

    The European Union demanded TikTok change its recommendations, saying repeated content may harm users’ well-being.

    BrazilG1, Tecnologia

    Quote
    A plataforma já está sob suspeita de ter cometido outras infrações a esse mesmo regulamento, incluindo o descumprimento de suas obrigações de proteção a menores.
  6. A Jev that can ask Opus for help gives Opus accuracy at 4x lower cost

    A test found that combining Jev with Claude Opus better judged which records belonged to the same person, with similar accuracy at over four times lower cost.

    Hacker News348 points · 77 comments

    Quote
    We tested Jev and GPT-6 Luna as judges for identity resolution. A hybrid of Jev and Claude Opus matched frontier accuracy at over 4x lower cost.
  7. Scientists just found dinosaur blood that survived 115 million years

    Scientists chemically identified traces of ancient blood in dinosaur bones that are 115 million years old.

    ScienceScienceDaily

    Quote
    Scientists have discovered chemically confirmed traces of ancient blood inside 115-million-year-old dinosaur bones, challenging assumptions about how long biological molecules can survive.
  8. WeLM-Image: A Unified Image Generation Model towards Visual Intelligence

    The team presents a model that creates and edits images from text and reference images, and reports strong results in tests.

    Papernot peer reviewed · 39 votes

    Quote
    We present WeLM-Image, a unified image foundation model that integrates multimodal understanding into text-to-image generation, instruction-guided editing, and multi-reference synthesis.
  9. An unexpected feedback loop may be driving mega heatwaves

    A study finds that dry soil, high pressure and extreme heat reinforce one another, intensifying heatwaves beyond climate model predictions.

    ScienceNew Scientist

    Quote
    Dry soil and high pressure can amplify extreme heat, helping to explain why regions like Europe are suffering record-shattering heatwaves much earlier than climate models predicted
  10. Deus Ex

    The piece revisits Deus Ex, highlighting its complex gameplay and questioning exaggerated claims about its predictions and philosophical meaning.

    Hacker News342 points · 177 comments

  11. We may have accidentally sent the strangest messages to 29 alien worlds

    Researchers suggest that early radio broadcasts from Earth may have reached 29 distant worlds and could have been heard by aliens.

    ScienceBBC Science Focus

    Quote
    Could our earliest radio broadcasts have been heard by alien worlds? Researchers suggest the waves have travelled over 100 light-years by now
  12. TailSteer: Trajectory-Tail Steering for Drift-Resistant Multi-turn Image Editing

    A study presents a training method that reduces unwanted visual changes across repeated image edits while preserving the model’s existing capabilities.

    Papernot peer reviewed · 37 votes

    Quote
    Strong single-turn editing performance does not guarantee reliable multi-turn interaction: as edited outputs are repeatedly reused, small unintended changes can accumulate into severe visual drift.
  13. Matt Damon’s ship from ‘The Odyssey’ needs your help

    Anyone who can get to Norway by spring can help repair the 114-foot vessel used in Christopher Nolan’s film.

    SciencePopular Science

    Quote
    If you can make it to Norway by spring, you could help fix the 114-foot-long vessel in Christopher Nolan’s blockbuster.
  14. Pwn.college

    pwn.college is a free Arizona State University platform for learning cybersecurity through hands-on exercises.

    Hacker News306 points · 53 comments

  15. EvoMemState: Evolving Memory State for Personal Agents

    A study proposes that digital assistants store personal facts, how they relate, and how to apply them when answering each request.

    Papernot peer reviewed · 20 votes

    Quote
    Long-term memory enables large language model (LLM) agents to provide continuous personalized assistance by retaining user information across interactions.
  16. RIP Cybertruck

    Tesla decided to stop producing the Cybertruck, an angular pickup that WIRED describes as unpopular and absurd.

    Hacker News301 points · 4 comments

    Quote
    The decision to halt production of Tesla’s polarizing polygonal pickup marks the end of an error.
  17. Towards Foundation Decision Models: Drex

    The team introduces four open models that choose among options or rate answers, aiming for faster, more consistent results.

    Papernot peer reviewed · 18 votes

    Quote
    Autoregressive approaches with large language models, such as long chain-of-thought reasoning, have become the default for new AI applications because they reach higher accuracy on a wide range of tasks.
  18. From Linux to Wikipedia: How Volunteers Build the Internet

    No summary: the source’s text could not be read.

    Hacker News301 points · 102 comments

  19. VisSearch-Bench: Evaluating Long Horizon Visual Search in Multimodal Agents

    Researchers created tasks to test searches combining images and online information; the best system fully solved only about a quarter of them.

    Papernot peer reviewed · 17 votes

    Quote
    Multimodal agents use web and image tools to answer questions beyond their internal knowledge. Some search tasks use images mainly to identify an initial entity, then rely on text retrieval.
  20. Gemini Omni 1.2 Flash

    Google’s Gemini Omni 1.2 Flash creates and edits videos up to 15 seconds long, with more consistent results and a 30% lower price.

    Hacker News299 points · 220 comments

    Quote
    Gemini Omni 1.2 Flash brings native 15-second video generation and editing, stronger consistency, and 30% lower pricing.
  21. Open-HumanScale: An Open Recipe for Scaling Human and Robot Experience for Embodied Foundation Models

    A study collected videos of people handling objects and used them to train robotic systems that adapt to new visual goals.

    Papernot peer reviewed · 15 votes

    Quote
    Egocentric human video offers a scalable source of manipulation priors for robot learning, yet translating raw data scale into generalizable robot policies demands rigorous data curation and training recipes.
This week’s papers

How this is made

A script picks the links by votes or by date. An AI summarizes each one using only the source’s text, and writes the day’s three lines from those summaries. Nobody reviews them before they go up, so check the original before citing.

Data in JSON · RSS

Sources: Hacker News, arXiv, Hugging Face, OpenRouter, Quanta Magazine, New Scientist, BBC Science Focus, Popular Science, Nature, Science News, Scientific American, MIT News, NASA, ScienceDaily, Agência FAPESP, Pesquisa FAPESP, Folha, Ciência, G1, Tecnologia, Tecnoblog, MIT Technology Review, WIRED, IEEE Spectrum, Ars Technica, The Verge, TechCrunch, The Register, Simon Willison, OpenAI, Google DeepMind, Google AI, Hugging Face.