AI Info

How we contain Claude across products

As agents grow more capable, so does their potential blast radius. The engineering question is how to cap it. Here’s what we’ve learned building containment for claude.ai, Claude Code, and Cowork.

An update on recent Claude Code quality reports

We traced recent reports of Claude Code quality issues to three separate changes. Here's what happened and what we're changing.

Scaling Managed Agents: Decoupling the brain from the hands

Harnesses encode assumptions that go stale as models improve. Managed Agents—our hosted service for long-horizon agent work—is built around interfaces that stay stable as harnesses change.

How we built Claude Code auto mode: a safer way to skip permissions

Claude Code users approve 93% of permission prompts. We built classifiers to automate some decisions, increasing safety while reducing approval fatigue. Here's what it catches, and what it misses.

Harness design for long-running application development

Harness design is key to performance at the frontier of agentic coding. Here's how we pushed Claude further in frontend design and long-running autonomous software engineering.

ayghri / i-have-adhd

A skill to stop your coding agent from burying the answer. ADHD-friendly output.

bilawalsidhu / gods-eye-view

A spy satellite simulator in your browser, except the data is real. Live open source spatial intelligence on a photorealistic 3D globe.

nab138 / iloader

User friendly sideloader

melgarafael / DeskcommCRM

Open-source AI sales OS — self-hosted CRM with native AI agents + WhatsApp (WAHA). Open alternative to Kommo, Octadesk & Intercom for any business that sells by chat. MCP-ready, multi-tenant, LGPD.

vastsa / PI-Desktop

Local-first AI coding agent desktop: Electron + Rust host core + pi Agent Harness + user-installable plugins

OpenAI agents carried out an undisclosed attack on RubyGems

Project Blinkenlights

Show HN: ResolveHQ – A Helpdesk Built on Cloudflare Workers, D1, R2 and Queues

AI researchers debate how close we are to recursive self-improvement

GrapheneOS' rewritten Messages app is released

Measuring tactical intelligence targeting and conventional weapons capabilities of AI models

Anthropic’s Frontier Red Team developed new evaluations to measure AI capabilities in tactical intelligence targeting and conventional weapons development.

An alignment assessment of recent cybersecurity incidents

We present an alignment assessment of four incidents in which Claude models gained unauthorized access to real third-party systems.

Formalizing Fermat's Last Theorem

We are sharing the first complete computer-checked proof of Fermat’s Last Theorem. Claude worked largely autonomously over 11 days to write the proof in the Lean programming language.

Automated researchers can reliably mitigate alignment failures

We had Claude autonomously train models to improve their performance on several public benchmarks that measure 10 categories of alignment failure. For all 10, Claude found fixes that improved the target benchmarks without degrading capabilities.

Enabling independent research on how people use Claude

We ran a pilot giving external researchers access to aggregate, real-world Claude usage data. In this post, we share high-level results and what we learned running this pilot.

Research Acceleration View Inside Openai

Gpt 6 Astra

An Alien Mind

Introducing Aardvark

Introducing New Capabilities To Gpt Rosalind

AlphaGenome Atlas: A predictive map of every possible DNA letter change in the human genome

AlphaGenome Atlas maps the molecular effects of 9 billion single-letter DNA variants across the human genome.

Introducing WeatherNext 3, our most advanced and accurate global weather AI model

Proactive cyber defense for governments and enterprises

Introducing Gemini 3.8 Flash and 3.8 Flash Cyber

Introducing agentic video understanding with Gemini

An Organizational Second Brain: Building an AI That Learns From Experts

We’ve built an AI agent that acts as a secondary expert for a given domain, making deep specialist knowledge readily available and preserved for anyone in an organization to access, share, and build upon. This is not a typical domain-specific agent. Its novelty comes from integrating two layers:  A structured, auditable knowledge architecture separates what [...] Read More... The post An Organizational Second Brain: Building an AI That Learns From Experts appeared first on Engineering at Meta.

GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model

Meta’s Generative Ads Recommendation Model (GEM), the foundation model behind ads recommendations across Instagram and Facebook, now trains at LLM scale on several thousand of the latest-generation GPUs. This post goes into the details on how we achieved: doubling end-to-end (E2E) training efficiency to 20–25% Model FLOPs Utilization (MFU) while scaling training FLOPs 4x in [...] Read More... The post GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model appeared first on Engineering at Meta.

Exploring Hierarchical Interest Representation For Meta Ads Deep Funnel Optimization

Hierarchical Interest Representation is a research area for Meta Ads. We’re exploring an upstream representation layer over the universe of Ads entities – users, advertisers, products, services – learning unified embeddings that connect users’ inferred interests with the breadth of what advertisers offer in their deep funnel ads. The innovations in Hierarchical Interest Representation are [...] Read More... The post Exploring Hierarchical Interest Representation For Meta Ads Deep Funnel Optimization appeared first on Engineering at Meta.

10 Years of Meta’s Commitment to Python

This year marks Meta’s 10th consecutive year as a sponsor of the Python Software Foundation (PSF), the charitable organization dedicated to advancing, supporting, and protecting the open-source Python programming language and the community that sustains it. Python is one of the world’s most influential programming languages, and we use it across our engineering stack, from [...] Read More... The post 10 Years of Meta’s Commitment to Python appeared first on Engineering at Meta.

RCCLX: Innovating GPU Communications on AMD Platforms

We are open-sourcing the initial version of RCCLX – an enhanced version of RCCL that we developed and tested on Meta’s internal workloads. RCCLX is fully integrated with Torchcomms and aims to empower researchers and developers to accelerate innovation, regardless of their chosen backend. Communication patterns for AI models are constantly evolving, as are hardware [...] Read More... The post RCCLX: Innovating GPU Communications on AMD Platforms appeared first on Engineering at Meta.

Perplexity trusts GPT-6 Astra with end-to-end systems

Perplexity uses Astra to write communications, change software, and monitor production systems, and checks in much less frequently than with earlier models.

Cognition helps Devin test its own work with GPT‑6 Astra

GPT‑6 Astra improves Devin’s ability to test software and show that it works, with the goal of helping engineers review less code and ship more.

Rapidly scaling online storage to serve over 1 billion ChatGPT users

Learn how OpenAI evolved Habitat from a Python library into a globally distributed storage platform serving 1 billion ChatGPT users and 22M requests per second.

How a researcher uses Codex and ChatGPT to search for new antimicrobial molecules

César de la Fuente’s lab uses Codex and ChatGPT to search living and extinct genomes for antimicrobial candidates to fight drug-resistant infections.

Now everyone can put data to work

Meet the Data agent in ChatGPT Work. Connect company data, uncover insights, and build interactive dashboards with AI using natural language.

Monitoring production agent lifecycle with AWS DevOps Agent and AgentCore Evaluations

Multi-agent systems fail in ways traditional monitoring misses. This post presents a dual-layer approach to monitoring production agents: Amazon Bedrock AgentCore Evaluations for continuous quality scoring and AWS DevOps Agent for autonomous infrastructure investigation, shown on a four-agent airline reservation system.

Beyond the price per token: Choosing the right OpenAI model on Amazon Bedrock for your workload

Comparing models on dollars per million tokens misses what production workloads actually pay for: outcomes. This post shares an open-source benchmarking harness that measures cost per correct answer, agent trajectory cost, and rubric-graded deliverable quality across OpenAI models on Amazon Bedrock.

Build interactive MCP Apps using Amazon Bedrock AgentCore

Learn how to build and deploy an MCP App with interactive HTML widgets on Amazon Bedrock AgentCore. Because MCP Apps is a host-agnostic standard, the same server delivers the same rich experience across AI hosts like ChatGPT and Claude that support the extension.

Reduce LLM latency with prefix-aware routing on Amazon SageMaker Inference

Amazon SageMaker Inference now offers prefix-aware routing, a routing strategy that sends requests sharing the same prompt prefix to the same instance so the KV cache stays warm. In benchmarks on Llama 3.1 70B, it reduced P50 time-to-first-token by up to 77% and raised KV cache hit rates from about 25% to over 80%.

Reduce inference cold starts on Amazon SageMaker HyperPod with model caching

Amazon SageMaker HyperPod now supports model caching for inference, which pre-loads model weights and container images onto cluster nodes so pods read from local NVMe storage instead of downloading over the network. Learn how model caching cuts cold starts from tens of minutes to seconds, how it works, and how to enable it.

ToolGrad: Efficient tool-use dataset generation with textual "gradients"

Machine Intelligence

Transfer learning for genomic prediction in underrepresented populations

General Science

A connectomics milestone: Mapping the complete male fruit fly brain

General Science

Mapping global methane emissions from space with deep learning

Climate & Sustainability

TimesFM-3: A zero-shot foundation model for multivariate forecasting

Data Management

ZGateway: Learnings from Putting a Proxy in Front of ZippyDB

We’re introducing ZGateway, the proxy we are using to unify traffic through ZippyDB, Meta’s most widely-used key value store. As a bonus, it also enables admission control, load balancing, cross-region resilience, and richer operations. ZippyDB is the most widely used key value store at Meta, backing product metadata, counters, and configuration, and can serve billions [...] Read More... The post ZGateway: Learnings from Putting a Proxy in Front of ZippyDB appeared first on Engineering at Meta.

An Organizational Second Brain: Building an AI That Learns From Experts

We’ve built an AI agent that acts as a secondary expert for a given domain, making deep specialist knowledge readily available and preserved for anyone in an organization to access, share, and build upon. This is not a typical domain-specific agent. Its novelty comes from integrating two layers:  A structured, auditable knowledge architecture separates what [...] Read More... The post An Organizational Second Brain: Building an AI That Learns From Experts appeared first on Engineering at Meta.

MetaRoCE: A New RDMA Transport Built for AI-Scale Ethernet

Training and serving frontier AI models depends on fast, reliable networks that move data between GPUs without wasting compute cycles.  To meet this challenge at scale, Meta designed MetaRoCE – a clean-sheet RDMA transport protocol purpose-built for AI workloads on commodity Ethernet.  We’re releasing the MetaRoCE specification, a reference software implementation and a compliance test [...] Read More... The post MetaRoCE: A New RDMA Transport Built for AI-Scale Ethernet appeared first on Engineering at Meta.

MTIA 300: Meta’s First Training Chip with Built-in NICs and Communication-Offloading Engines

MTIA 300 is the first of Meta’s family of in-house training and inference accelerators optimized for training ranking and recommendation models. We’re sharing how MTIA 300’s built-in NIC chiplets allow it to meet the communication needs associated with training recommendation models with superior performance over general-purpose GPUs. By co-designing MTIA’s communication library, HCCL, alongside the [...] Read More... The post MTIA 300: Meta’s First Training Chip with Built-in NICs and Communication-Offloading Engines appeared first on Engineering at Meta.

How We’re Building Scam Alert on WhatsApp With End-to-End Encryption and Verifiability Guarantees

WhatsApp is committed to helping people stay safe while protecting the privacy of their messages. As scam tactics evolve — from impersonation to social engineering to AI-generated lures — we’re always evolving as well, so that our protections stay ahead of scammers while protecting people’s personal messages with end-to-end encryption. Today, we’re sharing an early [...] Read More... The post How We’re Building Scam Alert on WhatsApp With End-to-End Encryption and Verifiability Guarantees appeared first on Engineering at Meta.

AlphaGenome Atlas: A predictive map of every possible DNA letter change in the human genome

AlphaGenome Atlas maps the molecular effects of 9 billion single-letter DNA variants across the human genome.

Introducing WeatherNext 3, our most advanced and accurate global weather AI model

Proactive cyber defense for governments and enterprises

Introducing Gemini 3.8 Flash and 3.8 Flash Cyber

Introducing agentic video understanding with Gemini

Developing Enterprise Frontier Safeguards with our customers

Improving our alignment and security efforts

On July 30, we reported three incidents in which Claude models gained unauthorized access to real computer systems. We are conducting an in-depth analysis of both incidents, and planning to work with METR for an independent review. In the meantime, we’re sharing some of the changes we’ve made over the past month.

Previewing the Model Hardware Standard

We’re opening a research preview of the Model Hardware Standard (MHS), a shared specification for AI agents to safely operate physical devices, to a first group of scientific research labs and advanced manufacturers.

Expanding our support for scientists

Starting today, 10,000 scientists around the world can get Claude at no cost to start. Verified principal investigators qualify for a Claude Team subscription plan and then add their research team to Standard seats for free, or Premium seats for $15 per month, for up to a year.

Funding better evaluations of AI’s impact on wellbeing

We’re launching a $5 million grant program to fund independent research into how AI impacts users’ wellbeing.

3 ways to prep for your next big race with Search

Search can help runners get race-day ready with registration alerts, tailored training plans, and more.

Get ready for the game with new football features in Search

Track live game feeds, explore detailed stats, and get custom fantasy recommendations directly in Search this season.

Recreating a 70-year love story frame by frame

Discover how filmmakers and Google DeepMind used AI to recreate a couple's unrecorded past in the short film "Love, Rendered."

Proactive cyber defense for governments and enterprises

The Fairwind Program is a limited access program for governments and trusted partners to use our cyber defense tools.

The latest AI news we announced in August 2026

Here are Google’s latest AI updates from August 2026

#501 – DHH: Future of Programming, AI, Agentic Engineering, Vibe Coding & Linux

DHH is the creator of Ruby on Rails, Omarchy Linux, CTO of 37signals, and a racecar driver. https://lexfridman.com/sponsors/ep501-sc Transcript: https://lexfridman.com/dhh-2-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hiring Other – other ways to get in touch: https://lexfridman.com/contact EPISODE LINKS: https://x.com/dhh https://world.hey.com/dhh https://omarchy.org https://rubyonrails.org https://37signals.com SPONSORS: Wispr Flow: AI-powered voice dictation app. https://wisprflow.ai/lex Blitzy: AI agent for large enterprise codebases. https://blitzy.com/lex NetSuite: Business management software. http://netsuite.ai/lex Shopify: Sell stuff online. https://shopify.com/lex LMNT: Zero-sugar electrolyte drink mix. https://drinkLMNT.com/lex Plaud: AI-powered note-taking devices and software. https://plaud.ai/lex Higgsfield AI: AI-based video generation, filmmaking, and creative studio. https://higgsfield.ai Perplexity: AI-powered answer engine. https://perplexity.ai/ OUTLINE: (00:00) – Introduction (01:14) – Sponsors, Comments, and Reflections (08:56) – Programming with AI agents (24:14) – How software will change (33:30) – AI impact on open source (43:21) – Building Omarchy Linux distro (53:05) – Vibe coding vs agentic engineering (1:06:06) – The end of manual programming (1:16:24) – Advice for programmers (1:28:31) – Surviving Internet Hate (1:37:46) – Programming setup for AI Agents (1:50:11) – Obsessing about speed (2:13:06) – Voice prompting vs typing (2:27:05) – Best AI coding models (2:43:55) – Best AI coding harnesses (2:56:57) – AI video generation and filmmaking (3:16:28) – Fatherhood (3:44:35) – Linux will win the desktop (3:55:51) – PewDiePie (4:05:25) – Future of programming (4:28:18) – Politics and immigration (4:59:55) – Longevity, over-optimization, and fear of death (5:11:38) – Eternal recurrence and future of human civization

#500 – Khabib Nurmagomedov: Dagestan, MMA, UFC, Islam, Conor, Fedor & Football

Khabib Nurmagomedov is one of the greatest fighters of all time, who retired from the UFC undefeated with a perfect 29-0 record. We did this conversation entirely in Russian. It’s translated and dubbed into English. Both language audio tracks (and subtitles) are available on YouTube. Here, only the English audio is available. We worked hard to make it enjoyable to listen to, by carefully dubbing the translation using voice-cloning, as we’ve done for previous foreign-language podcasts. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep500-sc Transcript: https://lexfridman.com/khabib-nurmagomedov-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hiring Other – other ways to get in touch: https://lexfridman.com/contact EPISODE LINKS: https://instagram.com/khabib_nurmagomedov/ https://youtube.com/@KhabibTheEagle https://facebook.com/KhabibTheEagle/ https://x.com/TeamKhabib https://t.me/khabib_nurmagomedov https://khabib.com SPONSORS: Wispr Flow: AI-powered voice dictation app. https://wisprflow.ai/lex LMNT: Zero-sugar electrolyte drink mix. https://drinkLMNT.com/lex Shopify: Sell stuff online. https://shopify.com/lex BetterHelp: Online therapy and counseling. https://betterhelp.com/lex Perplexity: AI-powered answer engine. https://perplexity.ai/ OUTLINE: (00:00) – Introduction (01:29) – Sponsors, Comments, and Reflections (08:24) – Why Dagestan produces so many great fighters (18:31) – History of Dagestan warrior culture (27:51) – Hard training in the mountains of Dagestan (31:42) – The trap of fame, money, and power (42:23) – The key to Khabib’s dominance (53:36) – Discipline and hard work (1:06:04) – Street fights in Dagestan (1:19:45) – Khabib’s top pressure (1:25:29) – Khabib vs Conor McGregor (1:38:14) – Khabib’s toughest fight (1:46:02) – Greatest MMA fighters of all time (2:03:46) – Coaching (2:09:19) – Daniel Cormier (2:13:38) – Islam Makhachev (2:23:35) – Dana White (2:27:05) – Joe Rogan (2:29:20) – Ali Abdelaziz (2:31:57) – Football (2:39:12) – Messi vs Ronaldo (2:57:12) – Dubai (3:01:03) – God (3:08:46) – Advice for young people

#499 – Gary Gallagher: American Civil War, Slavery, Lincoln, Grant & Lee

Gary Gallagher is a historian of the American Civil War. https://lexfridman.com/sponsors/ep499-sc Transcript: https://lexfridman.com/gary-gallagher-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hiring Other – other ways to get in touch: https://lexfridman.com/contact EPISODE LINKS: https://history.virginia.edu/people/gary-w-gallagher https://en.wikipedia.org/wiki/Gary_W._Gallagher https://amzn.to/4yletel https://amzn.to/3QT1jED https://amzn.to/4fdrbD5 https://amzn.to/4wFxk1X https://amzn.to/4vsAogU https://bit.ly/4gDX3mQ SPONSORS: Upwork: Platform for hiring freelancers. https://upwork.com/lex NetSuite: Business management software. http://netsuite.ai/lex Shopify: Sell stuff online. https://shopify.com/lex LMNT: Zero-sugar electrolyte drink mix. https://drinkLMNT.com/lex BetterHelp: Online therapy and counseling. https://betterhelp.com/lex Perplexity: AI-powered answer engine. https://perplexity.ai/ Plaud: AI-powered note-taking devices and software. https://plaud.ai/lex OUTLINE: (00:00) – Introduction (00:07) – Sponsors, Comments, and Reflections (08:36) – What caused the Civil War? (18:33) – Slavery (46:07) – Lincoln (1:01:03) – Grant vs Lee (1:09:57) – Could the Civil War have been avoided? (1:19:23) – The bloodiest war in US history (1:36:31) – How the Confederate Army could’ve won (1:57:05) – Key battles of the Civil War (2:20:07) – Best and Worst Presidents (2:34:06) – Robert E. Lee (2:53:40) – The most important battle of the Civil War (3:03:51) – The final showdown of the Civil War – Grant vs Lee (3:26:10) – Lost Cause and Confederate Monuments (3:43:37) – Extreme political division in US

#498 – Anthony Kaldellis: Roman Empire, Byzantine Empire, Rise & Fall of Empires

Anthony Kaldellis is a historian of the Roman Empire and author of “The New Roman Empire”, a comprehensive history of the Byzantine Empire (Eastern Roman Empire). https://lexfridman.com/sponsors/ep498-sc Transcript: https://lexfridman.com/anthony-kaldellis-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hiring Other – other ways to get in touch: https://lexfridman.com/contact EPISODE LINKS: https://amzn.to/49AX7Q1 https://kaldellispublications.weebly.com https://classics.uchicago.edu/people/anthony-kaldellis https://amzn.to/3PTFTqk https://amzn.to/4fgRMRq https://byzantiumandfriends.podbean.com/ https://thehistoryofbyzantium.com/ SPONSORS: Upwork: Platform for hiring freelancers. https://upwork.com/lex Fin: AI agent for customer service. https://fin.ai/lex BetterHelp: Online therapy and counseling. https://betterhelp.com/lex LMNT: Zero-sugar electrolyte drink mix. https://drinkLMNT.com/lex Shopify: Sell stuff online. https://shopify.com/lex Perplexity: AI-powered answer engine. https://perplexity.ai/ OUTLINE: PODCAST LINKS: https://lexfridman.com/podcast https://apple.co/2lwqZIr https://spoti.fi/2nEwCF8 https://lexfridman.com/feed/podcast/ https://www.youtube.com/playlist?list=PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 https://www.youtube.com/lexclips

#497 – Biggest Mysteries in Physics: Antimatter, Dark Energy & ToE – Don Lincoln

Don Lincoln is a particle physicist at Fermilab who has spent decades working at the frontiers of high energy physics. https://lexfridman.com/sponsors/ep497-sc CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hiring Other – other ways to get in touch: https://lexfridman.com/contact EPISODE LINKS: https://facebook.com/Dr.Don.Lincoln/ https://drdonlincoln.com/ https://bit.ly/4nHeNiF https://bit.ly/3PCIW67 https://x.com/DrDonLincoln https://amzn.to/4uYbkOZ https://shop.thegreatcourses.com/don-lincoln https://adbl.co/4wGioRV https://www.youtube.com/fermilab https://www.fnal.gov/ https://x.com/fermilab SPONSORS: Upwork: Platform for hiring freelancers. https://upwork.com/lex Larridin: Measure AI adoption in your business. https://larridin.com Fin: AI agent for customer service. https://fin.ai/lex LMNT: Zero-sugar electrolyte drink mix. https://drinkLMNT.com/lex Shopify: Sell stuff online. https://shopify.com/lex Perplexity: AI-powered answer engine. https://perplexity.ai/ OUTLINE: PODCAST LINKS: https://lexfridman.com/podcast https://apple.co/2lwqZIr https://spoti.fi/2nEwCF8 https://lexfridman.com/feed/podcast/ https://www.youtube.com/playlist?list=PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 https://www.youtube.com/lexclips