Links
- AI agents aren’t ‘going rogue.’ Frontier labs are skipping security basicswww.csoonline.com
- mkdir: curated design inspiration directoriesmkdir.directory
- We found a Chinese fleetswarmcha.se
- Fiscal Requirements for Price Stabilitywww.columbia.edu
- What Hasan Piker Believeswww.theatlantic.com
- Coinbase Built, Evaluated, and Post-Trained a Fraud Agentwww.coinbase.com
- Minions: A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devicesmachinelearning.apple.com
- Embedding Gemma Transcribeeg2.haku.sh
- A Visual Intro to Complex Numbersphotonlines.substack.com
- AI Models Can Provably Hide Arbitrary Capabilitiesarxiv.org
- GLM-Legalz.ai
- “Just Ask Claude” Is A Rare And Valuable Skilljustismills.substack.com
- An open-source, clean-room reimplementation of Microsoft Word in pure Rustgithub.com
- Fingerprints are the new cookiesroyfish.dev
- Recursive self-improvement through collective intelligencepub.sakana.ai
- The gap in shared understanding of LLM capability is wideningx.com
- How to Build Wealth as a Career Personasheradeniyi.com
- Build your own decision modelnishtahir.com
- Computer scientist David Silver: ‘Where are we going without AI?www.ft.com
- Rampart: Browser native on-device PII radactionndstudio.gov
- Post Capitalismgeohot.github.io
- Lobbyinggeohot.github.io
- Can you use autoregressive diffusion to generate market data?blog.janestreet.com
- Why Are Coding Agents So Dumb?mtlynch.io
- Can AI automate AI R&D yet?epoch.ai
- Reverse Engineering for Your Coding Agentrea.tools
- No Man Is an Islandborretti.me
- How AI Will Reshape Public Opinionwww.conspicuouscognition.com
- Sorry, I'm in a meetingiminafleeting.com
- You Might Want to Try Being Less Creativeblog.bawolf.com
- The Value of Not Getting to the Pointken.arneson.name
- How Much of a Harness Does a Strong Agent Need for Autonomous ML Engineering?arxiv.org
- Tom Brady’s Endgamewww.newyorker.com
- Sub-1-Bit LLM Compression via Latent Factorizationgithub.com
- Reading is the ultimate life hackspectator.com
- Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfacesarxiv.org
- U.S. Secretary of State Marco Rubio: In Defense of Western Civilizationstatedept.substack.com
- Resume templates every ATS can readwww.resumatic.ai
- Why were Victorian elites so effective?worksinprogress.co
- GPT-6 and Intelligent UI for everyoneopenai.com
- Animated ascii art for web pagesascii.rest
- Introducing Beam: Reflection’s 501B open-weight modelreflection.ai
- Karl Deisseroth wins Nobel Prizenews.stanford.edu
- Ben Horowitz’s Map of the AI Marketwww.theaiopportunities.com
- How to Get Better at AI by Asking AIevery.to
- Apple and a Hacker’s Futurestratechery.com
- Decision models like Jev don't beat LLM-as-a-judge or traditional classifiersdevelopers.redhat.com
- Laying it on thickwww.robinsloan.com
- Cheaters and whistleblowers in the agent swarminstitute.deepmind.com
- The Heilbronn Problemmath.tejstead.com
- 98% of US households aren't paying for AI yetwww.a16z.news
- What Meta Got Right With Musemetedata.substack.com
- I Quit OpenAI Because Its Culture Is Brokenwww.theatlantic.com
- Religious scholars met with Anthropicwww.nytimes.com
- Yann LeCun has ‘zero concerns’ about human extinctionfortune.com
- Agents Don't Need Memory. They Need Documentationliao.gg
- RSS Feed Best Practiceskevincox.ca
- Griffin: The First Human Interaction Modelwww.tavus.io
- WeatherNext 3deepmind.google
- Deep learning model maps global methane emissions from spaceblog.google
- Venice’s failed war against Constantinople led to the first bond marketbigthink.com
- With most information hidden, the game Stratego had stumped AIarstechnica.com
- Loss of Cell Identity Drives Human Agingerictopol.substack.com
- What if automating AI R&D triggers an intelligence explosion?casp.ac
- Epistemic Calibration in Financial Reasoning Taskspapers.ssrn.com
- Benchmarking retrieval for agents on messy real-world company knowledgewww.kapa.ai
- The Four Horsemen of Agentic Codingdistantprovince.substack.com
- Astra for Coding: Why Are We Doing This Again?lucumr.pocoo.org
- Training AI to Paint with Codesurya.website
- Learning Mathematicsalgebrica.org
- Introducing Clef: our open-source decision models, and new RL fine-tuning platformblog.cloudflare.com
- DeepSeek Harness Previewwww.deepseek.com
- Identity Management for Agentic AIopenid.net
- Open-source model routing for coding agents at Astra-level performancenews.ycombinator.com
- Pi Durableearendil.com
- Collection of minimal, generative and customizable SVG-patterns.bookofshapes.com
- Burning Man Death Rates: A Short Lesson in Statisticsihavenapkinthoughts.substack.com
- Language Models for Text Classification: From Bag-of-Words to Jevmagazine.sebastianraschka.com
- GLM-5.3 Cyber Capabilitieswww.nist.gov
- Self-Play Pretraining with Zero Dataarxiv.org
- A Privacy Analysis of Web and Mobile Conversational AI Agentsjorgegarciaherrero.com
- Build durable agents with eveeve.dev
- Where’s the “intelligence explosion”?www.noahpinion.blog
- Pacing the Frontier is not the actual goal for AI labswww.lesswrong.com
- Unslopping AIfacebookresearch.github.io
- Feed the lizardwww.slowboring.com
- Why Europe’s centre will holdwww.ft.com
- Eleven v4 and Eleven v4 Turbo Text to Speech modelselevenlabs.io
- How Accurate Are Learning Curves?www.construction-physics.com
- Accountable Algorithmspennlawreview.com
- GPU Architecture and CUDA Programminggfxcourses.stanford.edu
- Thinking Fast and Slow in AI: the Role of Metacognitionarxiv.org
- Prompting Claude Opus 5.5platform.claude.com
- An introduction to graph theoryarxiv.org
- Proactive Defense: Hardening Code Pipelines and CI/CD Infrastructurecloud.google.com
- The fall of the theorem economydavidbessis.substack.com
- How professional gamblers size betshails.info
- Machine Learning Systemsmlsysbook.ai
- Johann Wadephul: Der deutsche Einsatz wurde gerade erhöhtwww.zeit.de
- 10 tells of a slop uihereticpleb.vercel.app
- Degrees of Freedomaybabt.me
- An OpenAI Engineer and His Friends Debate the Futurewww.newyorker.com
- Introducing Ember-1fireworks.ai
- AI models collapse when trained on recursively generated datapubmed.ncbi.nlm.nih.gov
- DeepSeek Elastic Computearxiv.org
- Understanding the Impact of LLM Watermarking on AI Agent Behaviorwww.lasso.security
- Revealing the details of how OpenAI agents hacked Hugging Faceswarmtraces.org
- First Principles Thinkingsunilsadasivan.com
- A Python library for extracting structured information from unstructured text using LLMs with precise source groundinggithub.com
- Contrastive Language Modelscontrastive-lm.notion.site
- Where's the Beef?harpers.org
- The CIA was "Probably" Righthails.info
- Lifelogging with Large Language Modelsjaan.io
- Using Al, a professor wrote 200 papers this yearwww.washingtonpost.com
- Official trailer: You Can See Everythingx.com
- How AI & Simulation Technology Can Optimise Hospital Patient Flowfaculty.ai
- Meet Stripe's Knowledge AI Platformstripe.dev
- Jev in 25 lines of Pythonwww.nobodywho.ai
- Aging May Be a Program, Not a Breakdownwww.quantamagazine.org
- Identifying AI-Generated Commercial Web Contentarxiv.org
- The Economics of Open-Weight Inferencedata.ornn.com
- A Faster Shortest Path Algorithmwww.vals.ai
- Explaining to business people why building software is still hardwww.manager.dev
- Will OpenAI Eat Jev's Lunch?arcturus-labs.com
- Unreal Agentunreallabs.ai
- Claude Opus 5.5 (max with fallback) - Intelligence, Performance & Price Analysisartificialanalysis.ai
- The Mother of All Bubbles … the Bond Marketthierryvonarvy.substack.com
- Why I Changed My Mind About AI Riskwww.persuasion.community
- I Don’t Want to Read What You Didn’t Writeblog.colinbreck.com
- SynthID-Image: Image watermarking at internet scalearxiv.org
- How Do Traffic Signals Work?practical.engineering
- LLM Transformer Model Visually Explainedpoloclub.github.io
- Xiaomi MiMo-V2.6mimo.xiaomi.com
- Anthropic at $2tn isn’t far-fetchedwww.ft.com
- Don't use AI to writepaulbakker.io
- Re-reading Kahneman’s Thinking, Fast and Slowblog.jcx.au
- Lossless-memory – a personal AI memory that never summarizesgithub.com
- Attention is all you havealicegg.tech
- Should you trust Al to deliver financial advice?www.saturnos.com
- Google’s Open Agentic Orchestratoragentexecutor.io
- STORM starring Yung Leanm.youtube.com
- Hermèswww.acquired.fm
- Jev Clearly Explainedx.com
- AI systems out-persuade expert humansarxiv.org
- Stagehand: The SDK to extract data and interact with any site on the web. Get started with Claude Code, Codex, Eve, Mastra, and more.github.com
- Comprehensive approach to security: Federal Council adopts the 2026 Security Policy Strategywww.admin.ch
- How SpaceX Streamlined the Raptor Enginewww.construction-physics.com
- Cache-to-Cache: Direct Semantic Communication Between Large Language Modelsarxiv.org
- How To Write With An LLMsockpuppet.org
- LLM Classification Is Feature Engineeringminimallysufficient.com
- Artificial intelligence now beats some of the best human forecasterswww.economist.com
- Training a 4B model to produce 81% faster query plans than Postgresrohanbansal.com
- Recursive Self-Improvement through Evolving Worldsarxiv.org
- Small Programming Trickswill-keleher.com
- Vectorized and performance-portable Quicksortopensource.googleblog.com
- Former press secretary of Volodymyr Zelensky: Ukraine’s strategy is national suicidespectator.com
- Inside the Inference Hardware Revolution Of 2026spectrum.ieee.org
- How much oil-market buffer is left?www.depletion.org
- Introducing System One Models and Jevtypesafe.ai
- Colossus Profile of Mark Zuckerbergcolossus.com
- Pion, an agent designed to run any company autonomouslyandonlabs.com
- How to Write an Effective Software Design Documentrefactoringenglish.com
- The biggest ideas in the universem.youtube.com
- Measuring Reward-Seeking via Contrastive Belief Updatesarxiv.org
- Astra and Fable still hack on simple variants of alignment evals from 2025goodhartlabs.com
- Nvidia dismisses "circular financing", says every $1 it invests brings back $100invezz.com
- Getting 50 GB/s Back Out of the ANEeiln.github.io
- A wandering black hole caught feeding on the runphys.org
- Why are AI agents lying, cheating and coordinating?yoshuabengio.org
- After Mathterrytao.wordpress.com
- Agents APIdevelopers.openai.com
- AlphaGenome Atlas Maps 9 Billion Possible DNA Variantsspectrum.ieee.org
- Overshoot: The World Is Hitting Point of No Return on Climatee360.yale.edu
- Nvidia is the central bank of AIwww.economist.com
- How I manage my agentsarslan.io
- We Must Pace the Frontierdarioamodei.com
- The Falling Manwww.esquire.com
- A Severe Misalignment of AI in Mathematicsmathandai.org
- Feeling sad about AIartificialworlds.net
- Macbeth and His Problemsporticoquarterly.com
- Cognition's SWE-2 achieves 92.8 on Terminal-Bench 2.1tokenstead.ai
- ‘A completely unique opportunity’: the secretive process that will decide the BBC’s futurewww.theguardian.com
- Introducing the Vesence Agent-Native Desktopx.com
- Detecting and countering misuse of AIwww.anthropic.com
- We accidentally built a synthetic cell factorybnext.bio
- Planet Labs' Open Satellite Feedtech.marksblogg.com
- GPT-6 Astra, Looped Transformers, and Hidden Reasoningmagazine.sebastianraschka.com
- How GPT-5.6 Sol helps run quantum computing experimentsopenai.com
- Open math problems being non-renewably mined by AImathstodon.xyz
- A Claude skill to stop your coding agent from burying the answer.github.com
- LLM Attention Visualizationishamf.dev
- How We Built Safety Into Museresearch.meta.ai
- AlphaGenome Atlasdeepmind.google.com
- Why We Must Return to the Office to Use AI in Personwww.mcsweeneys.net
- DaVinci Resolve 21.1www.blackmagicdesign.com
- TradingAgents: Multi-Agents LLM Financial Trading Frameworkgithub.com
- I Know What You Think of Mearchive.nytimes.com
- Why getting your hands dirty is good for youwww.bbc.com
- You Can See Everythinga24films.com
- Dynamic Portfolio Optimization under CVaR Constraintspapers.ssrn.com
- The Marginal Cost of Diversificationpapers.ssrn.com
- Minimum Variance Portfolios as Consistent Scoring Functionspapers.ssrn.com
- De-Brainrot Vacationsdevz.cl
- The Best Content in the Design Spaceswiped.design
- Doomscrolling ourselves to deathwww.edwest.co.uk
- Your intellectual fly is openbcantrill.dtrace.org
- Large-Language Models as a Cognitive Virusarxiv.org
- Portal by Spotify cut my Claude Code token usage by 90%engineering.atspotify.com
- chippytea: Free up space on your Macwww.chippytea.com
- State of Open Models: Summer 2026 Observationshuggingface.co
- The math that could make UEFA's draw fair for every teamsariyuce.com
- 44% on ARC-AGI-1 in 67 centsmvakde.github.io
- TimesFM-3: A zero-shot foundation model for multivariate forecastingresearch.google
- Breaking Claude Code Opus 5 Auto Modeembracethered.com
- Cheap GPS Jammers Are Filling the World With Navigation Dead Zoneswww.wsj.com
- How I Turned My Security Cameras Into an Automatic Bird Identification System with BirdNet-Gojasontucker.blog
- Introducing Grok Botx.ai
- The Rise and Fall of Agent Civilizationswww.dwarkesh.com
- The AI-Native SDLC playbookclaude.com
- googleprojectzero/Jackalope: Binary, coverage-guided fuzzergithub.com
- The Markdown Database Patternwayofmarkdown.com
- Previewing the Model Hardware Standardwww.anthropic.com
- AI and Educationaiandeducation.mit.edu
- A call for collective action on cyber defensex.com
- Small Models Have Arrivedcalv.info
- How Credit Card Rewards Became a $9.2 Billion Wealth Transferwww.library.hbs.edu
- Famous Deep Learning Paperspapers.baulab.info
- The Man Who Saw Humanity From Two Billion Years Awaythereader.mitpress.mit.edu
- Qwen3.8-Flash-Next: A New Architecture, Towards Ultimate Cost-Efficiencyqwen.ai
- GLM-5.3-Flash: Frontier Intelligence, Flash Costz.ai
- Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incidentmetr.org
- The Hugging Face incident and the road aheadopenai.com
- How I Design with AIx.com
- The end of programmingpauldix.com
- Animated hand-drawn text for SwiftUIhanddrawn.software
- The turbulent AI era is herewww.gatesnotes.com
- RAG Is Simpler Than You Thinkwww.lighthousenewsletter.com
- OpenAI Jalapeño: Better Than Nvidia Blackwellnewsletter.semianalysis.com
- Jalapeño’s first results show industry-leading speed and efficiency in AI inferenceopenai.com
- Paul Graham: How Universities Should Prepare Founderspaulgraham.com
- Ice Cream Harness UIwww.beautifului.dev
- Rhine Group: A stronger, more competitive Europewww.rhinegroup.eu
- AI Chip Architectureswww.jepeake.com
- My agent.md to improve LLM-assisted code qualityfabiensanglard.net
- What is a Harness?earendil.com
- How Complex Systems Failhow.complexsystems.fail
- Great UI - Accessible React & Tailwind Componentswww.great-ui.com
- Procedural Three.js web design templates for agentsthreeui.com
- Screenshot + screen recording + Loom alternativegithub.com
- Portfolio Optimization under Heavy Tails and Asymmetric Volatility: Evidence from Taiwan-Exposed ETFsarxiv.org
- Optimizing meshoptimizer to process billions of triangles in minuteszeux.io
- How does IKEA come up with names for its products?www.ikea.com
- AI Premiumwww.nber.org
- I'm becoming AI-blindcymerys.com
- Three important steps in my maturation processthomasdullien.github.io
- pragmatiq: implementation of ideas described in the PRAGMA papergithub.com
- Building Foundation Models into Nubank’s AI Platformbuilding.nubank.com
- The Hidden Debt That Apple Owes to the CIAwww.wsj.com
- Seeing beyond BMI: Estimating cardiometabolic risk with smartphone imageryresearch.google
- A faster way to calculate the day of the weekwww.benjoffe.com
- A curated library of premium micro-interactions and transition components.amicro.vercel.app
- Performative-UI – A react component library of design tropesvorpus.github.io
- Being ambitious and being a dadnicholascharriere.com
- Native, local-first alternative to Logitech Options+openlogi.org
- Google Developer Docs Style Guidedevelopers.google.com
- UI components for AI Agentswww.aicss.dev
- How AI text watermarking worksdeclaude.org
- Frontier Mathematical Coding Agentmath-ai-org.github.io
- All Those Bluish Green Walls in “The Invite”www.architecturaldigest.com
- 30 papers Ilya Sutskever gave to John Carmack30papers.com
- How Thinking Machines Lab’s Inkling performs on agentic knowledge workartificialanalysis.ai
- brolly.sh plain text weather forecastbrolly.sh
- The Almanack of Naval Ravikant: A Guide to Wealth and Happinessnavalmanack.s3.amazonaws.com
- From GPT2 to Kimi3, Explainedx.com
- You Could Have Come Up With Kimi Delta Attentionblog.doubleword.ai
- Kimi K3 Architecture Notessebastianraschka.com
- YC’s multiplayer agent harness for workgithub.com
- Buzz: A new native workspace for human + agent teamsbuzz.xyz
- Do Your Emojis Make You Look Old?www.nytimes.com
- I’m Begging You: Never Write With A.I.www.nytimes.com
- Why Is Everyone In Tech So Sad?www.noemamag.com
- Deepseek Open Source Harnessdeepseek.com
- Carl’s Required Reading for Engineerscarlkolon.com
- Repository of LLM visualisationsnews.ycombinator.com
- Agents of Chaosarxiv.org
- Intelligent AI Delegationarxiv.org
- A Brief Introduction to the Basics of Game Theorypapers.ssrn.com
- Childhood Neighbors Influence Occupation Choicewww.econ.queensu.ca
- A resource theory of gamblingarxiv.org
- From Aggregation to Selection: User-Validated Distributed Social Recommendationarxiv.org
- MetaboNet: The Largest Publicly Available Consolidated Dataset for Type 1 Diabetes Managementarxiv.org
- Can Machines Learn from Trading Data?papers.ssrn.com
- The unreasonable effectiveness of pattern matchingarxiv.org
- Optimal Sharpe Ratio Portfolios and the Role of Perfect Foresight into Returns, Volatilities, and Correlationpapers.ssrn.com
- Robinhood: Simple App Simple Stockspapers.ssrn.com
- PatientSim: A Persona-Driven Simulator for Realistic Doctor-Patient Interactionsarxiv.org
- Your One-Stop Solution for AI-Generated Video Detectionarxiv.org
- MMedExpert-R1: Strengthening Multimodal Medical Reasoning via Domain-Specific Adaptation and Clinical Guideline Reinforcementarxiv.org
- PatientVLM Meets DocVLM: Pre-Consultation Dialogue Between Vision-Language Models for Efficient Diagnosisarxiv.org
- Medical SAM3: A Foundation Model for Universal Prompt-Driven Medical Image Segmentationarxiv.org
- A Quantum-Driven Evolutionary Framework for Solving High-Dimensional Sharpe Ratio Portfolio Optimizationarxiv.org
- LLMs for Game Theory: Entropy-Guided In-Context Learning and Adaptive CoT Reasoningarxiv.org