What we're reading, in the order we're reading it.
Anthony and Harry's working stream — links, charts, tweets, and short takes. The unprocessed inputs behind the briefings.
Indeed and Glassdoor are cutting 1,300 jobs as they go all in on AI
As part of the move, San Francisco–based Glassdoor will be absorbed into Indeed, and both companies will see senior leadership changes.
Moonshot AI’s Kimi K2 outperforms GPT-4 in key benchmarks
Chinese AI startup Moonshot releases open-source Kimi K2 model that outperforms OpenAI and Anthropic on coding tasks with breakthrough agentic capabilities and competitive pricing.
A new paradigm for AI: How ‘thinking as optimization’ leads to better general-purpose models
A new AI model learns to “think” longer on hard problems, achieving more robust reasoning and better generalization to novel, unseen tasks.
AI therapy bots fuel delusions and give dangerous advice, Stanford study finds
Popular chatbots serve as poor replacements for human therapists, but study authors call for nuance.
Anthropic May Never Catch OpenAI. But It’s Already 40% as Big.
Anthropic is ~40% as big as OpenAI by ARR, despite its consumer app Claude being just 5% the size of ChatGPT. The AI race isn’t just about who has the...
Vitalik’s Response to AI 2027
Daniel notes: This is a linkpost for Vitalik’s post. I’ve copied the text below so that I can mark it up with comments. • … …
OpenAI’s Windsurf deal is off — and its CEO is going to Google
Key researchers from the AI coding startup are also heading to Google.
A language model built for the public good
ETH Zurich and EPFL will release a large language model (LLM) developed on public infrastructure. Trained on the “Alps” supercomputer at the Swiss National Supercomputing Centre (CSCS), the new LLM...
New AI system uncovers hidden cell subtypes, boosts precision medicine
MIT researchers led a collaborative project resulting in a new AI system (CellLENS) that reveals hidden patterns in cell behavior within tissues and builds a comprehensive digital profile of individual...
How Trump’s megabill could slow AI progress in US
The elimination of federal renewable energy tax credits in the “big, beautiful” law has implications for the global AI race, and the US may be on track to lose to...
Goldman Sachs is piloting its first autonomous coder in major AI milestone for Wall Street
The arrival of agentic AI at Goldman Sachs, in which programs don’t just help with tasks but execute complex jobs, signals a shift on Wall Street.
OpenAI Is About to Release an AI Web Browser
OpenAI is reportedly just weeks away from releasing an AI web browser with a “native” ChatGPT-like interface.
OpenAI’s Sam Altman says your computer isn’t built for AI — so it’s creating something entirely new
The computers of the future won’t look the same
Elon Musk xAI Unveils Grok 4, Calls it ‘World’s Smartest AI’
New model boasts expanded capabilities as previous model sparks controversy
GitHub – cactus-compute/cactus: Cross-platform framework for deploying LLM/VLM/TTS models locally on smartphones.
Cross-platform framework for deploying LLM/VLM/TTS models locally on smartphones. – cactus-compute/cactus
Google for Developers Blog – News about Web, Mobile, AI and Cloud
Learn about GenAI Processors, a new open-source Python library for building powerful and flexible Gemini applications with streamlined, real-time multimodal processing.
Indonesia on Track to Achieve Sovereign AI Goals With NVIDIA, Cisco and IOH
Leaders from these companies reveal the nation’s new AI Center of Excellence, supported by the Indonesia government, that will foster localized AI research, nurture local talent and drive innovation with...
Apple AI model flags health conditions with up to 92% accuracy
Apple researchers developed an AI model trained on behavioral data collected from wearables, and it performed surprisingly well.
Metacognition and Self-Modeling in LLMs
Do frontier LLMs know what they know or know what they’re going to say? • An interim research report …
The Latest 20VC + SaaStr: Meta’s $1B Talent Raid, the End of Triple-Triple-Double-Double, and Why AI Laggards Are Getting Left Behind
Three Big Things This Week: Daniel Gross & Nat Friedman walked away from a $1.1B fund (already 4x’d) to join Meta – signaling the biggest talent war in Silicon Valley...
How to run SWE-bench Verified in one hour on one machine
We are releasing a public registry of optimized Docker images for SWE-bench. This allows us to run SWE-bench Verified in 62 minutes on a single GitHub actions VM.
Perplexity’s Comet AI browser is hurtling toward Chrome – how to try it
The agentic browser aims to compete with Chrome and Safari as the AI browser race heats up. Here’s how it works.
Elon Musk introduced Grok 4 last night, calling it the ‘smartest AI in the world’ — what businesses need to know
Musk did not apologize nor did he accept responsibility for Grok’s antisemitic, sexually offensive, and conspiratorial remarks.
Perplexity launches Comet, an AI-powered web browser
Perplexity is launching its first AI-powered web browser, Comet, marking the company’s latest attempt to unseat Google Search.
Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity
METR released a new paper with very interesting results on developer productivity effects from AI. I have copied the blogpost accompanying that paper…
Everything tech giants will hate about the EU’s new AI rules
EU rules ask tech giants to publicly track how and when AI models go off the rails.
Google Launches AI Marketing Tools in India After Tax Repeal, Drives 11% Ad Revenue Growth
Google Launches AI Marketing Tools in India After Tax Repeal, Drives 11% Ad Revenue Growth
Genies Is Building The Visual Layer For LLMs
Genies partners with Unity to bring emotionally intelligent AI avatars into games, merging behavior, memory, and UGC tools for real-time interactive companions.
Transparency Framework Proposed for Frontier AI Models
Anthropic released a proposed transparency framework to keep large AI developers in check
Evaluating and monitoring for AI scheming
As AI models become more sophisticated, a key concern is the potential for “deceptive alignment” or “scheming”. This is the risk of an AI system beco…
Kaya Brings AI to Construction Supply Chains, Procurement
Supply chain management in construction, despite its complexities, has remained stubbornly resistant to technology upgrades.
Elon Musk xAI Importing Power Plant for New Data Center
Elon Musk has confirmed that his artificial intelligence company, xAI, is purchasing a power plant overseas and plans to ship it to the U.S. to power a new data center.
Can AI Replace Air Traffic Controllers to Reduce Airline Accidents?
Tests in London and Singapore could reveal whether AI can improve the safety of air travel
Can a $9bn deal sustain CoreWeave’s stunning growth?
The AI superstar faces competition and an over-reliance on big tech
Kimina-Prover: Applying Test-time RL Search on Large Formal Reasoning Models
A Blog post by Project-Numina on Hugging Face
ScreenEnv: Deploy your full stack Desktop Agent
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
We Sent 4,495 AI SDR Emails in 2 Weeks and Got The #1 Response Rate. Here’s How (It’s Not What You Think)”
After less than 30 days with our AI SDR, we became the highest response rate on our AI SDR platform. #1. Here’s what actually works — and what doesn’t. _______...
OpenAI set to launch new browser very soon to compete with Google Chrome — here’s what we know
Watch out Google, OpenAI is coming
EU finalizes General-Purpose AI Code of Practice for enterprises
The code of practice arrives just in time to help enterprises comply with the second wave of EU AI Act rules, which will enter force on August 2.