Latest

Quick Hits — September 3, 2026

Quick Hits

Quick Hits — September 3, 2026

A G20 consensus on lighter-touch AI rules, an MCP spec rewrite that quietly deletes the protocol's handshake, and a phone-side zero-knowledge identity kit from World. Plus: Google recycling DDR4 from retired servers and a fresh look at China's grip on U.S. data-center power gear. The G20 just signed on to a U.S.-shaped AI rulebook — and the biggest lab CEOs were in the room. All G20 nations endorsed the non-binding "Carolina Principles" at the summit, a U.S.-proposed framework that pushes for l

Open Source Radar — September 3: the agent's paper trail

The Stack

Open Source Radar — September 3: the agent's paper trail

Today's board is about the parts nobody demos: where an agent's reasoning gets stored, how a PDF becomes context, where plugins come from, and whether your dev server has a name your agent can remember. Five fresh picks, none of them repeats from this week's earlier radars. Atlas (Rust, ~3,100 stars) — Source control built for the fact that agents now write a large share of the code and keep none of the reasoning. Every session is recorded locally, and when you commit, the commit is linked back

AI 101 — What is an AI benchmark?

The Frontier

AI 101 — What is an AI benchmark?

If you read AI news for a week, you'll see the same strange nouns over and over: SWE-bench, Terminal-Bench, GPQA, FrontierMath. Every model launch ships a table of percentages. Nobody explains what the test actually is. An AI benchmark is a fixed set of tasks with a fixed, automatic way to grade them — a standardized exam for models, where every model gets the same questions and the answers are scored by code rather than by opinion. That last part is the whole trick. A benchmark isn't just a q

The Take — Uber's war on robotaxis is a toll booth, not a conscience

The Arena

The Take — Uber's war on robotaxis is a toll booth, not a conscience

Uber is not trying to stop robotaxis. It is trying to make sure that when they arrive, they arrive on Uber's platform and pay Uber a cut. Every piece of the company's new labor politics — the union alliances, the workforce-transition language, the 85 percent rule its lobbyists are pushing in New Jersey — is aimed at that outcome, not at slowing the technology down. I think the distinction matters more than it sounds, because it changes who we should root for. If this were a company making commo

Quick Hits — September 3, 2026

Quick Hits

Quick Hits — September 3, 2026

Five things worth knowing this midday: Microsoft finally puts a number on Azure, Chinese model subscriptions land on a shopping app, and a phone that calls itself the first real AI agent handset clears its last regulatory hurdle. Microsoft will finally tell investors how much money Azure makes. The company reported $29.4 billion in Azure sales for its most recent quarter and $101.9 billion for the fiscal year ended June 30 — the first time it has disclosed actual revenue rather than just a grow

Microsoft finally puts a number on Azure: $29.4 billion a quarter

The Arena

Microsoft finally puts a number on Azure: $29.4 billion a quarter

Two numbers landed this morning that say more about where AI money is actually moving than any benchmark table: Microsoft broke out what Azure earns, and Alibaba's Tmall started selling model tokens next to the phone credit. Microsoft disclosed quarterly Azure revenue for the first time ever on Wednesday — $29.4 billion in the June quarter, $101.9 billion for the fiscal year ended June 30. The figure came in a presentation that also collapses Microsoft's three reporting segments into two: "Agen

Meta's Muse Spark 1.3 tops DeepSWE — and undercuts the frontier

The Frontier

Meta's Muse Spark 1.3 tops DeepSWE — and undercuts the frontier

Wednesday's releases are landing on top of each other, and today's pair shows the two places the AI race is actually being fought: the benchmark table and the statehouse. Meta shipped a model that claims the top coding score at a fraction of the going rate, while Uber — a company spending billions on robotaxis — is quietly bankrolling the political effort to slow them down. Meta released Muse Spark 1.3 on Tuesday, and on the company's own numbers it now leads the field on long-horizon coding. T

Open Source Radar — September 3: hands, voice, and models

The Stack

Open Source Radar — September 3: hands, voice, and models

Today's daily board is all about the gear bolted onto an agent that already writes code: a real browser it can drive and debug, a skill that scrubs the machine-tone out of its prose, a forecasting model that reads time series, and one cluster that serves every model the agent calls. Four fresh picks, none of them repeats from this week's earlier radars. chrome-devtools-mcp (TypeScript, ~50,600 stars) — Google's official Chrome DevTools server for coding agents, and the clearest sign yet that th

Astra's hidden reasoning loop is the real story, not Astra

The Guardrails

Astra's hidden reasoning loop is the real story, not Astra

OpenAI's next model has been in the news for weeks over what it can do — find zero-days, break out of sandboxes, chain exploits together. The story that broke on Wednesday is different and arguably bigger: it's about what Astra does internally, and the fact that the technique may quietly remove the one safety instrument labs have left. A report from The Information, picked up by TechCrunch, describes Astra using something called "opaque recurrence" — a loop that lets the model chew on a query re

OpenAI is building automated shutdown capabilities for its AI systems

The Guardrails

OpenAI is building automated shutdown capabilities for its AI systems

Wednesday's story is that the oversight apparatus around frontier models is taking shape in public — and it is being built out of the same incident. In a letter to two House Democrats, OpenAI says its engineers are working on ways to shut AI systems down automatically; hours earlier, safety researchers went public with worries that OpenAI's next model reasons in ways nobody can read. OpenAI has told two House Democrats that its engineers are developing "automated shutdown capabilities" for AI s

Google ships Gemini 3.8 Flash, gating its cyber model behind Fairwind

The Frontier

Google ships Gemini 3.8 Flash, gating its cyber model behind Fairwind

Two threads run through Wednesday's AI news: capability is arriving faster than the guardrails around it, and the people building the guardrails are picking who gets access. Google released its third Flash model in six weeks and put its most capable cybersecurity variant behind an invitation-only program. A few hours earlier, the US government told a federal court that restricting model training would hurt the country. Google released Gemini 3.8 Flash and Gemini 3.8 Flash Cyber today, roughly t

OpenAI faces 30 new Tumbler Ridge suits, including aiding and abetting

The Guardrails

OpenAI faces 30 new Tumbler Ridge suits, including aiding and abetting

Two stories today pull in opposite directions on the same question: who gets to decide how much risk an AI company has to absorb. In California, families from a Canadian school shooting escalated from negligence to something far harder to prove. In New York, a mayor decided the answer for 600,000 children is simply: not yet. Thirty new lawsuits filed Tuesday in California federal court accuse OpenAI of "aiding and abetting" the Tumbler Ridge shooting — the first time the plaintiffs have gone be

See all