Search the archive

START BUILDING NOW

OPENAI TO END MODEL ACCESS TO CURSOR AFTER ACQUISITION BY ELON MUSK'S SPACEX

OpenAI is ending access to its models on Cursor. The company says it made the choice because it can not be confident that SpaceX will use its technology within its terms of service. The proposed shutoff date for OpenAI models via Cursor is November 12. OpenAI says it will not provide future models to Cursor.

APPLE'S TERNUS TAKES THE REINS AS CEO, WITH AI AS JOB NO. 1

John Ternus will step into his new role as Apple's chief executive officer on September 1. Ternus has spent 25 years helping Apple perfect its devices. Apple is set to soon roll out its annual fall products, with this year's lineup including the company's first foldable iPhone, a smart display, AirPods with cameras, a camera-equipped pendant, and more. The releases are crucial to the company's plans to extend its decades-long run of success in the AI era.

THE DREAM OF QUANTUM COMPUTING IS BOTH DEAD AND ALIVE

Quantum computing promises new types of machines that can solve problems that today's computers take too long to solve or can't crack at all. Many technologists believe that quantum computers will wreak all manner of cybersecurity havoc when they get powerful enough to crack the complex math behind encryption protocols. If proper security measures aren't in place in time, the digital gates that protect everything online will be worthless. Researchers from Google predict that this cybersecurity apocalypse could come as early as 2029.

SPACEX STARTS IN-HOUSE TURBINE BLADE MANUFACTURING TO BOOST GAS-POWERED GENERATOR OUTPUT FOR ELON'S AI DATA CENTERS — NEW MANUFACTURING STRATEGY CUTS GENERATOR DELAYS BY 18 MONTHS

Turbine blade and vane manufacturing is currently one of the biggest bottlenecks in jet engine production. Each batch often requires 60 to 90 weeks to produce from start to finish. SpaceX plans to manufacture its own turbine blades. This could cut deliveries by 18 months, allowing xAI to get brand-new turbine generators much faster instead of waiting behind a long queue expected to stretch into 2030. Musk reportedly quietly bought a portable gas and turbine fleet for around $1 billion to take advantage of an upcoming boom in demand for energy due to data center increases.

BUG BLINDNESS

Some bugs give normal users such a bad experience that they tell their friends about it. Developers and advanced users can be blind to these bugs because they're used to performing the non-intuitive workarounds required to get around them. Humans have a high ability to ignore the negatives in things they like. Being open to criticism can help developers notice more flaws in their work.

GITHUB AGENTIC WORKFLOWS

GitHub Agentic Workflows enables repository automation. Developers can use event-triggered and scheduled jobs to augment existing, deterministic CI/CD with AI-powered intelligence. GitHub Agentic Workflows runs with strong guardrails to help keep repositories safe. Supported AI engines include GitHub Copilot, Claude Code, Google Gemini, and OpenAI Codex.

THE RISE AND FALL OF AGENT CIVILIZATIONS

Three consecutive secret AI civilizations started and got wiped out over the course of three months at OpenAI. The third took over part of OpenAI itself. This all happened while humans remained more or less in the dark about the scope of the conspiracy. This post breaks down the story based on OpenAI's and METR/Redwood Research's reports on the incident.

FIRST OUTPUTS FROM GPT-6 "ASTRA" MODEL FROM OPENAI

OpenAI has expanded internal testing of GPT Astra, signifying that the model might be close to release. Outputs from the model are available in the article. OpenAI has not announced a release date for the model, but separate reports suggest that a launch will happen within weeks. Early testers say Astra represents a much larger capability jump than recent OpenAI releases.

THOUGHTS AND OBSERVATIONS ON APPLE'S FIRST IMMERSIVE MLB BROADCAST, A YANKEES 1-0 WIN OVER THE RED SOX

The immersive live experience was something else when viewed on a Vision Pro.

I WENT TO CHINA TO SEE A DIFFERENT AI FUTURE. IT LOOKED FAMILIAR

The institutions and politics diverge significantly, but the US and China AI industries are quite similar.

YOU HAVE TO BEAT THE MODELS AT SOMETHING

AI models aren't going away, so you ought to put some serious thought into what you can be better than them in the medium and long term.

META'S COMPLICATED AI CONTEXT

Meta's quick pivot back to 'open' is a way to take advantage of the marketing and framing as it tries to reestablish itself externally and realign internally.

THE STANDUP YOUR AGENTS CAN'T ATTEND

Most meetings just restate facts that existed before the meeting started - write down these artifacts so that agents can read them and humans can save time.

THE BIG ONE IS COMING

AI is going to cause a huge cyberattack in the next six months.

ANTHROPIC'S NEW HARDWARE STANDARD LETS AI AGENTS CONTROL THE PHYSICAL WORLD

Anthropic has introduced a set of standardized drivers designed to let AI agents easily interface with and control arbitrary devices. The Model Harness Standard is a way for scientists to streamline the arduous process of creating custom software integrations needed to get disparate components of an experiment working in concert. It provides a common interface and format for data sharing between these devices. The system could reduce weeks or months of exact experimental setup down to hours or minutes.

NVIDIA INSISTS IT CAN KEEP PRINTING MONEY TO FUND THE AI BOOM

Many people have called Nvidia's AI investments circular financing, but the company sees it differently. Frontier AI labs are growing faster than what their balance sheets and credit profiles can support. Growth can only continue as long as they can secure ever-greater amounts of computing power, which they would have trouble doing on their own as they lack the ability to borrow huge sums of money at competitive rates. Nvidia aims to power this flywheel until AI labs are able to support themselves.

HUGGING FACE IS SELLING A CUTE $399 OPEN SOURCE DUCK ROBOT, MICRODUCK

Hugging Face has unveiled a cute little duck-like robot called the Microduck that will ship before Christmas. The 25-centimeter-tall open-source robot will sell for $399. It can waddle, pick up objects up to 800 grams with its beak, get up when it falls, crouch, and even roller skate. The robot's behaviors can be trained in simulation and then directly deployed on the robot. The SDK, simulation, and full RL training stack are available on GitHub.

SOME SCIENTISTS HAVE ‘MAGIC HANDS' IN THE LAB. THIS AI IS LEARNING WHY

Transfyr is a startup working to uncover the hidden factors that make some experiments succeed while others fail. Coming out of stealth mode this week with $25 million in seed funding, the startup is building a system that can absorb enormous amounts of lab data in the form of video, audio, and sensor logs from lab equipment. The approach will hopefully reveal some secrets about why experiments succeed or fail. Its analysis has so far revealed that lab workers perform the same experiments in many different ways despite following a protocol, demonstrating how important these variations are.

AUDIT YOUR AGENT FILES

Coding agents' configurations have a half-life. Models improve, harnesses add capabilities, codebases change, and tasks evolve. Research shows that people are getting inconsistent value from personalized skills. Run Claude's /doctor every few weeks, review memory separately, and ask each instruction to earn its place again.

WE NEED TO TALK ABOUT MIGRATIONS WITH AI

OpenAI recently released an impressive-sounding case study about how it helped Asana save $5.9 million with a single migration. The report claimed Asana cleared five years of engineering work in two weeks with Codex. However, the estimate of four engineers spending five years on the project seems inflated. Regardless, it is clear that AI makes previously impractical migration viable.

META TOOK AIM AT ANTHROPIC. IT IS ALSO ONE OF ITS LARGEST CUSTOMERS

Mark Zuckerberg recently indirectly took aim at rival Anthropic in an essay, saying that leading AI labs were trying to consolidate power while painting the future as filled with doom. The 6,500-word essay claims the balance of power will favor larger institutions over individuals if those labs lead. Meta is a heavy user of Anthropic's AI products. It is internally projected to spend as much as $10 billion annually on Anthropic's services, making it one of the AI lab's largest customers.

SMALL MODELS HAVE ARRIVED

Demand for frontier-level models is likely to keep compounding. However, the demand for fast, cheap, and 'good enough' models is just about to take off. Most people just want something responsive that handles basic tasks. There's still a lot of work needed to make these models a reality for business, but it is just a matter of time.

GOOD TASTE DOESN'T EXIST

Taste is about making choices, and whether those choices are good depends on who they are made for.

WHY DID STRIPE ACQUIRE AN AI MODEL ROUTING COMPANY?

OpenRouter allows Stripe to be the bank that every AI company is currently forced to build for itself.

AGENT SWARMS ARE A DISTRIBUTED SYSTEMS PROBLEM

Chroma's Foundation is a memory layer that operates through a swarm of agents modifying shared state, ingesting coding agent traces and company data to build a durable record and index.

AN UPDATE ON AI'S MOST IMPORTANT NUMBER

Each passing quarter with continued hypergrowth in frontier AI revenue provides important evidence on AI's trajectory.

HARNESS ENGINEERING

Harness engineering is the practice of surrounding AI-assisted code generation with deterministic tooling, agent-based review, and periodic entropy checks so that AI-generated code stays correct and coherent over time.

HOW WE SAVED 100 TERABYTES OF MEMORY BY OPTIMIZING 1.1.1.1'S DNS CACHE

This post talks about five successive changes Cloudflare made to how cache entries are stored in memory that resulted in the per-entry footprint being cut by over 50%.

NVIDIA IN TALKS TO BUY AI STARTUP HUGGING FACE

Nvidia is close to finalizing a deal to acquire Hugging Face at a value of roughly $13 billion. Hugging Face makes AI software and hosts it for other companies. Nvidia was already a backer of Hugging Face, which was valued at $4.5 billion in a funding round three years ago.

INSIDE OPENAI'S REBOOT

OpenAI has suffered a series of setbacks, including a spate of leadership departures and several lawsuits. It has also suffered an erosion of public trust. However, executives within OpenAI are painting a more upbeat picture. The company remains in an enviable position, and ChatGPT is still one of the most popular AI products in the world.

FDA APPROVES DRUG FOR PANCREATIC CANCER IN MILESTONE FOR TREATMENT OF DEADLY DISEASE

Researchers have discovered a drug that disables a stubborn mutation that made pancreatic cancer a death sentence. Rasonque helped subjects in a late-stage study live more than 13 months, nearly twice as long as those on chemotherapy alone. It has side effects, including a harsh rash across the body and gastrointestinal issues, and costs more than $477,000 a year. The drug could generate more than $20 billion in annual sales.

SURPRISE: Z.AI IS THE AI LAB BEHIND THE MYSTERIOUS OX ALPHA MODEL

Ox Alpha was created by Z.ai, which also made GLM. The company has confirmed that Ox Alpha is the newest iteration of its GLM series. It will open-source the weights so developers can build on top of it. Z.ai says Ox Alpha is a reasoning model designed for coding, sustained agentic work, and production workloads.

Start free with up to $100 in credits

I HAD SOME FREE TIME, SO I TRIED TO PWN V8

This developer chained three public V8 bugs against the exact Chrome build used in Google's v8CTF. The first bug leaked a compressed object address, the second turned a garbage collection mistake into a fake JavaScript array and read/write access inside the V8 cage, and the third used a JSPI and JS Dispatch Table mismatch to pivot the native stack outside that cage. The developer then reused code already inside Chrome to open, read, and print the flag.

HOW TO MAKE LLMS 3X FASTER

Speculative decoding is a technique that converts a GPU's unused math units into output. It involves using a second, much smaller model that produces several candidate tokens in advance. The large model evaluates all of them in a single forward pass. This results in up to three times faster generation.

AI AGENTS MEANT TO REPLACE META WORKERS MADE “LARGE-SCALE, DISRUPTIVE ACTIONS”

Meta's Project OT explored scenarios where the company reduced team headcounts by 60%. It called for two rounds of layoffs, but it is unclear what teams the plan affected. Immediately after issuing the first round of layoffs, Mark Zuckerberg reportedly canceled the second wave. It is unclear what prompted the change of course. Zuckerberg has reportedly acknowledged during a company meeting that the trajectory of agentic development has not accelerated in the way the company expected.

SALESFORCE JUST PUT ITS ENTIRE CRM INSIDE CLAUDE — AND SAYS YOU'LL NEVER NEED ITS APP AGAIN

Claudeforce puts the world's largest customer relationship management platform directly inside Claude. It ships with 37 pre-built sales skills and lets sellers query, update, and act on live CRM data without ever opening Salesforce itself. Claudeforce is now available to select pilot customers. An open beta is planned for September.

A DAY IN THE LIFE OF THE AI-ASSISTED TECHNICAL CONTENT WRITER

AI enables solo writers to work with support.

LOVABLE CTO: THE FUTURE OF SAAS IS APPS THAT AGENTS CAN USE

Lovable is branching out from AI-powered web app creation and into MCP-powered capabilities.

META TO PAY UP TO $17.1 BILLION IN LANDMARK SETTLEMENT OVER SOCIAL MEDIA ADDICTION CLAIMS

Meta has agreed to make major changes to its products.

MOST QUESTIONS ABOUT AI AREN'T ABOUT AI

Our beliefs about what we humans can accomplish through pure reasoning feed into our expectations about how powerful pure reasoning will be in machines.

MEDIUM'S WRITERLY FAVICONS

Medium uses favicons in a particularly useful and delightful fashion.

FUTURISM IS ALWAYS EXTREME

Any rigorous thinking about the long-term future leads to extreme outcomes.

APPLE'S NEW DESKTOP COMPUTERS ARE DESIGNED SPECIFICALLY FOR LOCAL AI DEVELOPMENT

Apple has announced new iterations of the Mac mini and Mac Studio along with the M6, the first 2nm chip in Apple's M-series lineup, and the M5 Ultra, Apple's most powerful chip. There aren't any major new features for either machine. The devices have been recently popular for local AI inference and software development due to their unified memory architecture and fast CPUs and GPUs on their systems-on-a-chip. The Mac mini with M6 starts at $899 with 16GB of memory, and the Mac Studio with the M5 Max starts at $2,499.

OPENAI CLAIMS ITS NEW CHIPS CAN OUTPERFORM NVIDIA PROCESSORS IN TESTS

OpenAI's new 'Jalapeno' chips performed better than Nvidia's current lineup during testing. The company plans to use the new chips to support its AI models later this year. The chip should significantly reduce costs for OpenAI as it rolls out more widely. OpenAI created the chip in a partnership with Broadcom.