Alvaro Lopez Ortega / 2025-07-20 Briefing

Created Sun, 20 Jul 2025 16:15:40 +0000 Modified Wed, 23 Jul 2025 20:04:55 +0000
4204 Words

Today’s headlines highlight the launch of Prime-One Fiber ISP in Michigan, challenging Comcast with gigabit speeds and expansion plans, and Amazon Ring’s rollback of privacy reforms to expand police access and AI surveillance, raising civil liberties concerns.

▶️ Internet Infrastructure

Brothers Launch Prime-One Fiber ISP to Compete with Comcast in Saline Michigan

Brothers-in-law Herman and Baciu built Prime-One, a fiber ISP in Saline, Michigan, offering gigabit speeds at $80/month, serving 1,500 homes with plans to expand beyond Comcast’s limited coverage.

  • Samuel Herman and Alexander Baciu founded Prime-One, a fiber ISP in Saline, Michigan, to replace Comcast due to poor service and slow upload speeds.
  • Prime-One offers symmetrical gigabit fiber for $80/month, with plans up to 5Gbps, unlimited data, no contracts, and includes equipment and installation at no charge.
  • The network covers about 1,500 homes with 75 miles of underground fiber, aiming to serve 4,000 homes and expand further, competing primarily against Comcast and Frontier.

Minimizing Dependencies Reduces Risks by Evaluating Cost Criteria

Using minimal dependencies like TigerBeetle reduces supply chain risks and complexity; dependencies should be evaluated across five criteria to avoid unnecessary costs.

  • Dependencies incur costs such as learning time, breaking changes, and deployment complexity.
  • TigerBeetle, a financial database, follows a “zero dependencies” policy, relying solely on the Zig toolchain.
  • The author proposes a dependency evaluation framework based on five categories: Ubiquity, Stability, Depth, Ergonomics, and Watertightness.

Amazon Ring Reverts Privacy Reforms to Expand Police Access and AI Surveillance

Amazon Ring reverts previous privacy reforms, enhances police access, and adopts AI-driven surveillance, risking civil liberties and enabling mass surveillance in the US.

  • Amazon Ring, led by founder Jamie Siminoff, reintroduces police access to user footage and live streams, emphasizing an “AI first” approach.
  • The company plans to incorporate video analytics and face recognition technology, requiring employees to demonstrate AI usage for promotion.
  • Ring is rolling back reforms like end-to-end encryption and ending police partnerships, increasing mass surveillance and civil liberties threats.

▶️ Open Source

Blender Studio Launches Free Open-Source Game DOGWALK with Creative Assets

Blender Studio released DOGWALK, a free open-source game using Blender and Godot, featuring a paper-crafted environment, watercolor textures, and player-driven interactions.

  • Blender Studio released the free game DOGWALK on July 11, 2025, for Windows, macOS, and Linux, rated 4.9/5 stars from 43 ratings
  • The game features a miniature open world where players control a big, adorable dog helping a child decorate a snowman, with no fail states and player-driven choices
  • Built with Blender and Godot, assets include real-life paper crafted models, watercolor textures, and scanned materials, licensed under CC BY 4.0 and MIT License

Mozilla to Gather User Feedback for Firefox Improvements Through Surveys and AMA

Mozilla seeks community input via surveys and an upcoming AMA to guide Firefox development, focusing on performance, customization, privacy, and technical features across platforms.

  • Mozilla is conducting community surveys and planning a Firefox AMA to gather user feedback on features, development priorities, and animal symbols representing browsing styles.
  • Users express desires for improvements in performance, interface customization, tab management, split view, and privacy features across desktop and mobile versions.
  • Specific technical requests include API enhancements (File System, Periodic Background Sync, MIDI Clipboard), native ad filtering, better extension support, and privacy technologies like DNS over Oblivious HTTP and DNS4EU.

Open-Source 8-DOF Humanoid Hand Project for Reachy2 Under €200

The AmazingHand project offers open-source code and models for a low-cost, 8-DOF humanoid hand with integrated actuators, supporting serial or Arduino control, aimed at exploring humanoid hand capabilities.

  • The AmazingHand project provides code and models to control an 8-DOF humanoid robotic hand with 4 fingers, designed for Reachy2, weighing 400g, and costing less than €200.
  • The hand features 2 phalanxes per finger, all actuators inside the hand without cables, and uses parallel mechanisms driven by Feetech SCS0009 servos; control options include serial bus with Python or Arduino with TTL Linker.
  • Resources include BOM, CAD files, assembly guide, and demo scripts for calibration and basic operation, with detailed instructions for both control methods.

Gwern’s 2021-2025 Experimentation with Affordable Avant-Garde Perfumes

Gwern sampled and evaluated experimental perfumes in 2021–2025, confirming their artistic potential; key scents include Acqua di Sale and Kyoto Incense, with samples costing as little as $6.20.

  • Gwern explored avant-garde and experimental perfumes in 2025, emphasizing affordable sampling via LuckyScent with $6.20 sampler bottles containing 10–20 doses.
  • He purchased 39 samples in 2021 costing $153, and a second batch of 10 samples in 2024 for $59, including scents like Room 237, Asphalt Rainbow, and Molecule series.
  • Top selected perfumes: Acqua di Sale ($111 for 50ml) and Kyoto Incense, intended for long-term use; other favorites include Lampblack and Molecule 03.

Helix 25.07 Update Adds File Explorer, LSP Colors, and Advanced Parsing Features

Helix 25.07, released on July 15, 2025, introduces a file explorer, LSP document colors, a rewritten command parser with flags and expansions, and a new tree-house crate for advanced Tree-sitter parsing, injections, and syntax highlighting.

  • Helix 25.07 was released on July 15, 2025, with contributions from 195 developers
  • Introduces a file explorer accessible via <space>e, functioning as a picker for directory hierarchy
  • Adds LSP document colors feature, requesting color ranges from language servers like tailwindcss-language-server and displaying inline color swatches
  • Complete rewrite of command line argument parsing, supporting flags (--reverse, --no-format) and expansions (%{variable}, %sh{}), improving completion and escaping
  • Replaces Tree-sitter bindings with new crate tree-house, enhancing syntax parsing, highlighting, and injection handling, including nested and incremental injections
  • Implements robust injection management with a tree of layers, optimizing re-parsing and highlighting in complex nested scenarios
  • Improves handling of locals (e.g., function parameters) by tracking definitions at parse-time, maintaining highlights regardless of view
  • Enhances injection support for all features, enabling consistent cross-language syntax behavior within layered syntax trees

Self-Taught Engineers Thrive Through Grit and Practical Experimentation

Self-taught engineers excel through purposeful tinkering and grit, developing intuition and mastery by experimenting, debugging, and learning from real-world failures rather than relying solely on formal education.

  • Self-taught engineers often outperform formally educated counterparts due to purposeful tinkering and grit.
  • Formal education emphasizes proficiency over intuition, which is critical for real-time problem solving.
  • Examples include Linus Torvalds, Margaret Hamilton, and open-source maintainers who gained mastery through trial, error, and real-world experimentation.

AI Industry Shifts to Hiring Top Experts for Better Training Data

AI firms are replacing low-cost gig workers with high-paid industry experts to improve training data quality, driven by advanced models like GPT-3 and Gemini 2.5, with investments exceeding $15bn.

  • AI industry shifts from low-cost African and Asian gig workers to hiring top experts in biology, finance, and other fields for training data creation.
  • Companies like Scale AI, Turing, and Toloka are recruiting specialists to develop more sophisticated, reasoning-based models such as OpenAI’s GPT-3 and Google’s Gemini 2.5.
  • Major investments include Meta’s $15bn in Scale AI (June), Turing AI’s $111mn raise (March), and Toloka’s $72mn funding led by Bezos Expeditions (May).

▶️ Software Development

Fstrings.wtf Offers Python 3.13 F-String Behavior Quiz with Scoring and Source Code

The site fstrings.wtf offers a 26-question Python 3.13 f-string behavior quiz, including scoring, answer selection, and source code on GitHub.

  • The webpage hosts a Python f-string quiz reflecting Python 3.13 behavior, created by Armin Ronacher and inspired by jsdate.wtf
  • Contains 26 questions with scoring, answer options, and navigation features; source code available on GitHub
  • Provides user interface elements for quiz management, sharing, and issue reporting, with links to creator profiles and source repository

Python AI Project with Monorepo, FastAPI, Docker, and CI/CD Tools

The author adopted Python for AI-driven development, utilizing a monorepo with integrated tools like uv, ruff, ty, pytest, Pydantic, FastAPI, and Docker, to build production-ready, scalable applications.

  • The author has been coding in Python for 6 months, primarily driven by AI opportunities and its ecosystem.
  • Uses a monorepo project structure with directories for backend (project-api) and frontend (project-ui), including CI/CD, documentation, and deployment files.
  • Employs tools such as uv for dependency management, ruff for linting, ty for static typing, pytest for testing, Pydantic for data validation, MkDocs for documentation, FastAPI for API development, and Dataclasses for data modeling.
  • Utilizes GitHub Actions for CI workflows, Dependabot for dependency updates, Gitleaks for secret scanning, and pre-commit hooks for code quality.
  • Implements Docker and Docker Compose for containerized deployment, with detailed docker-compose.yml and Dockerfile configurations for multi-service orchestration.

▶️ Management and Leadership

Lottocracy: Ensuring Meritocracy and Reducing Corruption Through Random Selection

Random selection (lottocracy) is essential for meritocratic stability, reducing corruption, and fostering diversity, as demonstrated by historical governance models and proposed multi-layered, transparent lottery systems.

  • Argues that stable meritocratic institutions require random selection (lottocracy) to prevent corruption and gaming of selection criteria
  • Highlights historical examples: jury systems, Venice councils, Athens’ random administrative roles, modern citizens’ assemblies
  • Proposes practical implementations: random oversight boards, random candidate selection, stratified sampling for committees, randomized auditing, nested lotteries for scalability

XMLUI Launches React-Based Visual Basic-Inspired Web Components

XMLUI, launched by Jon Udell, adapts Visual Basic component principles to web development with React-based components, enabling non-expert developers to build reactive, themeable apps with minimal deployment.

  • XMLUI, announced on July 18, 2025, by Jon Udell, brings Visual Basic-like component modeling to the modern web using React and XML markup.
  • Provides a suite of reusable components, including Select, DataSource, Table, and custom user-defined components, enabling reactive, themeable apps without deep React or CSS knowledge.
  • Demonstrates a minimal deployment footprint with a static HTML file, a JavaScript bundle, and XML UI files, supporting hosting on any static webserver or AWS S3.

Effective Backup Strategies: Ensuring Data Security and Consistency

A comprehensive backup system requires strategic planning, snapshot use for consistency, and choosing between full disk or file-based backups, with a focus on security, speed, and external storage.

  • Emphasizes the importance of a well-planned backup strategy before scripting, focusing on risk assessment, data protection, and storage considerations.
  • Discusses the advantages and disadvantages of full disk versus individual file backups, highlighting the need for snapshots to ensure data consistency.
  • Recommends using native filesystem snapshots (e.g., BTRFS, ZFS), LVM snapshots, or tools like DattoBD for reliable, consistent backups; advocates for push or pull architecture based on security needs.

AI Tools Slow Experienced Developers by 19% by Disrupting Mental Models

AI tools slow experienced open-source developers by 19% due to interference with mental model formation, as they rely on internal understanding that current AI cannot replicate effectively.

  • A July 2025 study found AI tools slow experienced open-source developers by 19% on familiar codebases, contrary to their belief of increased speed
  • Developers overestimate AI productivity gains, believing AI speeds them up by 24%, and maintain this belief even after experiencing slower task completion
  • The slowdown is linked to the theory by Peter Naur that programming is about forming mental models; AI tools hinder this process by lacking access to developers’ internal mental representations

Boost Coding Quality with Mental Proof Techniques for Faster, Correct Code

Practicing mental proofs—such as monotonicity, invariants, and induction—while coding enhances correctness, speed, and code quality by making reasoning about complex systems more intuitive.

  • Advocates writing small proofs mentally during coding to improve speed and accuracy
  • Highlights reasoning techniques: monotonicity, pre- and post-conditions, invariants, isolation, induction
  • Emphasizes structuring code for proof-affinity to enhance correctness and maintainability

The Limitations of Scaling Reliable Production AI Agents

Building over a dozen production AI systems, the author highlights exponential error compounding, quadratic token costs, and engineering challenges, arguing that reliable, scalable agents require bounded design and human oversight.

  • Building 12+ production AI agent systems across development, DevOps, and data operations reveals fundamental mathematical limitations.
  • Error rates compound exponentially in multi-step workflows, making 95% per-step reliability insufficient for 20+ steps; 20 steps yield only 36% success.
  • Context window costs scale quadratically, making conversational agents economically infeasible at scale; a 100-turn conversation can cost $50-$100 in tokens.
  • Successful production agents rely on bounded, verifiable operations with human oversight, not on fully autonomous, multi-step workflows.
  • Tool design is critical; effective feedback interfaces, partial failure handling, and structured responses are essential for reliability.
  • Integration with complex, legacy enterprise systems involves traditional systems programming, not just AI orchestration.
  • Practical, reliable agents operate within strict boundaries, with humans maintaining control over critical decisions and error recovery.
  • Predictions indicate that fully autonomous startups will face economic and reliability challenges, while domain-specific, constrained tools will succeed.
  • Building reliable AI agents requires clear boundaries, failure handling, cost management, and reliance on traditional engineering for execution and error handling.

OpenAI Enhances ChatGPT Agent Security After Red Team Finds Critical Vulnerabilities

OpenAI’s red team identified seven vulnerabilities through 110 attacks, leading to security enhancements like 100% traffic monitoring, operational restrictions, and classification of ChatGPT Agent as “High capability” for biological risks.

  • OpenAI introduced ChatGPT Agent, enabling autonomous login, email, file management, and web browsing for paying subscribers.
  • A red team of 16 PhD security researchers tested the system with 110 attack attempts, revealing seven universal exploits and critical vulnerabilities.
  • Post-testing, OpenAI implemented real-time traffic monitoring, multi-layer inspection, and strict operational restrictions, achieving 95% defense against visual browser attacks and enhanced biological and chemical safeguards.

Scott Werner Introduces a Four-Document System for AI-Driven Software Development

Scott Werner’s July 2025 article details a self-developed four-document system for AI-assisted programming, highlighting collective experimentation and the evolving nature of expertise in a rapidly changing AI landscape.

  • Scott Werner released a personal experiment with a four-document system for AI-driven software development in July 2025
  • The system includes: Architecture Overview, Technical Considerations, Workflow Process, and Story Breakdown, created organically during project work
  • Werner emphasizes the fluidity of expertise in AI development, describing current work as collective improvisation and experimentation

iCounter Secures $30M in Series A to Combat AI-Driven Cyber Threats

iCounter, a cyber risk intelligence firm utilizing AI for threat detection, raised $30 million in Series A funding to combat AI-driven targeted attacks on large enterprises.

  • iCounter raised $30 million in Series A funding led by SYN Ventures, announced on July 16, 2025
  • The company uses AI to detect and deflect highly customized cyber threats targeting enterprises
  • The funding will enable iCounter to exit stealth mode from inside Apollo Information Systems, where CEO John Watters is also executive chairman

From Pay Cut to Dream Job: Dawn Choo’s Journey to Data Science at Instagram

Dawn Choo shifted from finance to tech with a 40% pay cut, eventually becoming a data scientist at Instagram after multiple applications and career pivots, demonstrating long-term strategic risk-taking.

  • Dawn Choo took a 40% pay cut to transition from a quant role at Bank of America to a business analyst role at Amazon in 2017.
  • The pay cut significantly impacted her living situation, but she viewed it as a strategic step toward her goal of working at Meta.
  • After two years at Amazon, she was promoted to a business intelligence engineer and eventually secured her dream job as a data scientist at Instagram, applying to Meta about seven times over several years.

CZI Uses GPU Power to Attract Researchers Over Compensation

Priscilla Chan stated CZI leverages around 1,000 GPUs to attract researchers, focusing on compute power over compensation, aligning with Zuckerberg’s strategy to prioritize AI and scientific research.

  • Priscilla Chan highlighted access to approximately 1,000 GPUs at CZI as a key recruitment advantage
  • CZI plans to expand its GPU cluster to support biology research and attract top talent
  • Chan acknowledged CZI cannot compete with tech companies on compensation, emphasizing compute power as a strategic draw

US Plans to Revise H-1B Visa Lottery to Favor Skilled Applicants

US DHS and USCIS plan to revise the H-1B visa lottery system, potentially prioritizing skilled applicants, amid criticism that the current program favors foreign workers over US graduates and facilitates offshoring.

  • DHS and USCIS filed a regulatory notice on July 17, 2025, to reevaluate the H-1B visa issuance process, including a proposed rule titled “Weighted Selection Process for Registrants and Petitioners Seeking To File Cap-Subject H-1B Petitions.”
  • The review could conclude within days or weeks, with the new rule expected to be published in the Federal Register, potentially favoring applicants meeting specific skill criteria over the current lottery system.
  • The current H-1B program reached its Fiscal 2026 cap on July 20, 2025, with about 600,000 H-1B workers in the US as of 2019, and is criticized for displacing US computer science graduates and enabling offshoring practices.

▶️ Technology

AI Models Share Universal Representations, Supporting the Platonic Representation Hypothesis

The article presents the PRH, asserting that AI models develop universal, convergent representations, supported by embedding mapping techniques and circuit similarities, with potential applications in decoding complex languages.

  • The article discusses the Platonic Representation Hypothesis (PRH), proposing that different AI models learn convergent, shared representations of the world.
  • Evidence includes successful unsupervised mapping between embedding spaces using cycle consistency methods like CycleGAN, and similar functionalities in mechanistic interpretability studies.
  • The hypothesis suggests that as models grow larger, their internal features and circuits become more aligned, enabling translation and inversion across models and modalities, with implications for decoding ancient texts and whale speech.

Advances in Fully Homomorphic Encryption Boost Privacy-Preserving Cloud Computing

FHE allows encrypted data processing in real-world systems, with performance improving 8x annually, potentially enabling privacy-preserving cloud computing, LLM inference, and smart contracts.

  • Fully Homomorphic Encryption (FHE) enables arbitrary computations on encrypted data without decryption, ensuring data privacy during processing.
  • Current FHE incurs 1,000x to 10,000x overhead compared to plaintext operations, with ciphertext sizes 40 to 1,000 times larger.
  • FHE algorithms are improving at approximately 8x speed annually, with recent breakthroughs promising higher throughput and lower latency.

aws/pgactive Extends PostgreSQL with Active-Active Replication Support

aws/pgactive is a PostgreSQL extension implementing active-active replication via logical replication, supporting PostgreSQL 10+ features, with the latest version 2.1.5 released on June 19, 2025.

  • The aws/pgactive repository is an extension for PostgreSQL enabling active-active replication topology.
  • It supports logical replication features introduced in PostgreSQL 10, with ongoing development for full active-active support.
  • The latest release is version 2.1.5, released on June 19, 2025, with 11 total releases and 6 contributors.

Understanding Asynchrony in Zig: Tasks, Concurrency, and Parallelism

Asynchrony in Zig allows tasks to run out of order without requiring concurrency, enabling synchronous code to coexist with asynchronous code without duplication or deadlocks.

  • Defines asynchrony as tasks running out of order while remaining correct, distinct from concurrency and parallelism
  • Illustrates that io.async in Zig does not imply concurrency; code can run in single-threaded blocking mode
  • Differentiates between asynchrony (task order flexibility), concurrency (progress of multiple tasks), and parallelism (simultaneous execution at hardware level)

Firefox 141 Brings WebGPU Support to Windows with Cross-Platform Plans

Firefox 141 ships WebGPU support on Windows, leveraging WGPU for cross-platform graphics API compatibility, with future plans for Mac, Linux, and Android support.

  • Firefox 141 enables WebGPU on Windows after years of development; planned support for Mac, Linux, and Android in upcoming months
  • WebGPU implementation based on WGPU, a Rust crate supporting Direct3D 12, Metal, Vulkan, and OpenGL
  • WebGPU is defined in W3C standards WebGPU and WGSL; available in Chrome since 2023 and expected in Safari 26 this fall

AI Datacenter Spending Could Reach 2% of US GDP by 2025

AI datacenter spending, potentially 2% of US GDP in 2025, is reshaping economic statistics, causing sector reallocation, and driving mass job losses before widespread AI deployment.

  • AI datacenter capex may reach approximately 2% of US GDP in 2025, contributing an estimated 0.7% to GDP growth
  • Nvidia’s datacenter sales (Q1 FY2026) totaled $39.1 billion, with ~99% AI-related, representing roughly 25-35% of datacenter capex
  • AI capex has grown at least 10x since 2022, surpassing peak telecom spending and approaching railroad boom levels as a percentage of GDP

Sankalp’s Two-Week Review of Claude Code for Efficient Workflow and Model Insights

Sankalp reports two weeks of Claude Code usage, highlighting workflow techniques, context management, and model performance, with detailed insights into features, command tips, and model comparisons.

  • Sankalp used Claude Code with a $20 subscription, primarily relying on Sonnet 4, on medium-sized Python and large Ruby+TypeScript codebases, totaling over 50 million tokens.
  • Upgraded to a $200 Claude Max subscription for near-unlimited access to Sonnet 4 and Opus 4, enabling extensive experimentation and workflow optimization.
  • Utilized Claude Code within Cursor for diff review, code editing, and context management, employing features like scratchpad, session resumption (/resume), and memory files (CLAUDE.md).

Community Drives Development of Small Premium Android Phone to Rival iPhone Mini

A community-led initiative aims to create a small Android phone under 6" with flagship specs, targeting 50,000+ buyers to influence OEMs, featuring 1080p OLED, Pixel 5 cameras, Snapdragon 8, and stock Android, priced at $700-800.

  • The project aims to develop a small premium Android phone with a sub-6" display, matching iPhone 13 Mini size, with specifications including 5.4" 1080p OLED, Pixel 5-level cameras, Snapdragon 8 processor, and stock Android OS, priced at $700-800.
  • Ideal features include a consistent bezel width, 8GB RAM, 128/256GB storage, 4 hours SOT, unlockable bootloader, NFC, and optional ruggedness, water resistance, wireless charging, and eSIM.
  • Over 41,000 supporters have signed up, emphasizing community-driven effort to pressure manufacturers like Google or Samsung to produce the device, as no current premium Android phones meet these criteria since Sony Xperia Compact’s discontinuation.

Anthropic Tightens Claude Code Limits Without Notice, Disrupting Heavy Users

Anthropic has unexpectedly tightened Claude Code usage limits without prior notice, causing confusion among heavy users on the $200 Max plan and amid broader network issues, impacting service reliability and user trust.

  • Anthropic restricted Claude Code usage limits starting Monday without prior notice, mainly affecting heavy users on the $200/month Max plan.
  • Users report “Claude usage limit reached” messages with no clear indication of limit changes, causing confusion and project delays.
  • The company confirmed issues with response times but declined further comment; network errors and overloads also occurred during the same period.
  • Tiered pricing promises higher usage limits for Max ($200/month), Pro, and free plans, but actual limits are variable and not explicitly guaranteed.
  • The lack of transparent communication has led to user distrust, especially among high-volume users who rely on the service for significant API calls.

OpenAI’s Reasoning LLM Wins Gold at International Math Olympiad

OpenAI’s latest experimental reasoning LLM attained gold medal-level performance at the IMO, marking a significant breakthrough in AI’s mathematical reasoning capabilities.

  • OpenAI’s experimental reasoning large language model (LLM) achieved gold medal-level performance on the International Math Olympiad (IMO)
  • This accomplishment addresses a longstanding grand challenge in AI
  • The announcement was made by Alexander Wei on X on July 19, 2025, with the post receiving 4.5 million views

Apple’s 2026 Foldable iPhone to Launch Unconventionally Alongside Upgraded M5 iPad Pro

Apple’s 2026 foldable iPhone will debut with an unconventional launch, alongside the M5 iPad Pro’s dual front cameras, amid industry competition from Samsung and Google.

  • The first foldable iPhone is expected to launch in 2026, marking a departure from typical Apple product launches
  • The device will feature a less conventional, “un-Apple-like” release strategy
  • The M5 iPad Pro is set to receive significant upgrades, including dual front cameras

Eric Schmidt: AI Is a New Industrial Era Worth Trillions

Eric Schmidt argues AI is an emerging industrial structure rather than a bubble, citing hardware demand and chip market stability, with AI valued at $189 billion in 2023 and expected to reach $4.8 trillion by 2033.

  • Eric Schmidt states AI is a “whole new industrial structure,” not a bubble
  • AI market valued at $189 billion in 2023, projected to reach $4.8 trillion by 2033
  • Cites Nvidia’s hardware sales and data center capacity as signs of market longevity

LIGO-Virgo-KAGRA Detect Record-Breaking 225-Solar-Mass Black Hole Merger

LIGO, Virgo, and KAGRA detected GW231123, the most massive black hole merger to date, with a 225-solar-mass black hole formed from two highly spinning black holes of 100 and 140 solar masses.

  • LIGO-Virgo-KAGRA (LVK) Collaboration detected the most massive black hole merger, GW231123, on November 23, 2023, producing a final black hole ~225 solar masses.
  • The merging black holes had individual masses of approximately 100 and 140 solar masses, with the event observed during the fourth LVK observing run.
  • The black holes exhibited rapid spins near the theoretical maximum allowed by Einstein’s general relativity, challenging existing stellar evolution models.

Meta Opposes EU AI Code Overreach, Citing Growth Concerns

Meta refuses to sign the EU’s AI code of practice, claiming it overreaches and will stunt AI development; the rules aim to improve transparency and safety under the AI Act.

  • Meta Platforms declined to sign the European Union’s AI code of practice, citing it as an overreach that will hinder growth, according to global affairs chief Joel Kaplan.
  • The EU’s AI guidelines, aimed to enhance transparency and safety, are set to take effect next month and are designed to comply with the AI Act enacted last year.
  • Several companies, including ASML Holding and Airbus, have called for delaying the AI code, while OpenAI has committed to signing it; Meta opposes the regulations, claiming they introduce legal uncertainties and overreach.

Scientists Discover New “Altermagnets” That Could Revolutionize Spintronics and Quantum Tech

Researchers identified “altermagnets,” a new magnet class with unique spin-symmetry properties, potentially advancing spintronics and quantum computing technologies, first discovered in nearly 100 years.

  • Researchers discovered the first new magnet type in nearly a century, termed “altermagnets,” predicted by Libor Šmejkal in 2022 inspired by M. C. Escher’s artwork.
  • Traditional magnets are classified as ferromagnets and antiferromagnets; altermagnets exhibit a distinct spin-symmetry-breaking magnetic order, enabling unique electronic properties.
  • Altermagnets could facilitate development of novel spintronic devices and quantum computers due to their unconventional magnetic behavior.