Latest

Opus 5 cracks a second Enigma message — seven remain unsolved

The Frontier

Opus 5 cracks a second Enigma message — seven remain unsolved

Frontier models are quietly finishing the work Alan Turing never lived to see done — and the count of unsolved wartime ciphers is now low enough to fit on one hand. Claude Opus 5 has broken a second long-unsolved Enigma message — a German Army dispatch from 31 July 1941 that sat unread in the archive for 85 years — making it the second frontier-model cryptanalysis win validated in as many weeks. Cryptologist Frode Weierud published the write-up this week after Jack Willis contacted him on 21 Se

Appeals court upholds Pentagon's Anthropic blacklist on a 2-1 vote

The Guardrails

Appeals court upholds Pentagon's Anthropic blacklist on a 2-1 vote

Anthropic lost the appeal it needed most on Friday. The same day, the price tag on the government's other AI gatekeeping job leaked — and it dwarfs anything Congress has budgeted. A federal appeals court in Washington upheld the Pentagon's blacklisting of Anthropic in a 2-1 decision on Friday, rejecting the company's argument that the designation was arbitrary, unauthorized and unconstitutional. Circuit Judges Gregory Katsas and Neomi Rao found the Defense Department "had ample support for its

The Take — Medicare's denial pilot has the burden of proof backwards

The Guardrails

The Take — Medicare's denial pilot has the burden of proof backwards

Medicare's AI prior-authorization pilot is being argued about as if the scandal were the algorithm or the bounty. I think the real defect is smaller, plainer, and more fixable than either: the program never has to prove a denial was correct, and the only party who can force the question is the patient. Flip that burden of proof — put the reviewer before the patient rather than after — and most of this fight evaporates. Start with the money, because that's where the argument keeps getting stuck.

New York City wants auditors, a kill switch and a paid whistleblower

The Guardrails

New York City wants auditors, a kill switch and a paid whistleblower

Washington has spent a decade holding AI hearings and passing almost nothing. New York City Council now wants to see whether a city can do what Congress won't — starting with the systems its own residents are already using, sold by the five companies with the biggest offices in town. New York City Council Speaker Julie Menin unveiled a package of AI bills on Friday that would make it unlawful to market, sell or deploy an AI system in the city without third-party validation — and require that ev

Tesla's Optimus line is up tenfold — the hands are the bottleneck

The Arena

Tesla's Optimus line is up tenfold — the hands are the bottleneck

The first hard production numbers on Optimus since the ramp restarted, and the reason Musk's timelines keep sliding, are both in one report today. Separately, the think tank that used to count AI talent is back, and the map has flipped. Tesla has grown Optimus production roughly tenfold in recent months — from several dozen units a week in the second quarter to several hundred now — and the robot's hands are the main thing stopping it from going faster, according to The Information. The report

Skild's robot brain learned soccer by playing itself for 140 years

The Frontier

Skild's robot brain learned soccer by playing itself for 140 years

Skild AI says its S1 robotics model accumulated more than 140 years of simulated soccer inside NVIDIA's Isaac Sim, playing against recent copies of itself, and then transferred to a physical humanoid that played a real match. The published result, dated September 23, contains no success rate, no compute budget, no wall-clock training time and no comparison against a non-self-play baseline. What it does contain is a video of a robot in a blue-and-white jersey shielding a ball from a person, which

Japan's FSA puts AI data center loans under scrutiny

The Guardrails

Japan's FSA puts AI data center loans under scrutiny

Japan's financial regulator has started looking past the AI data centers and at the balance sheets paying for them. The Financial Services Agency named project finance — data centers included — a priority area for monitoring in its policy priorities for the 2026 fiscal year, and officials say the near-term focus is how the biggest banks and life insurers underwrite and then watch those exposures. Japan's lenders have become some of the largest creditors to the global AI buildout, largely becaus

DOJ joins Musk's appeal against the EU's €120 million X fine

The Guardrails

DOJ joins Musk's appeal against the EU's €120 million X fine

Washington has spent a year complaining about European tech enforcement. On Thursday it became a party to a lawsuit over it. Separately, the one European company the AI buildout cannot do without says its home market has stopped buying. The US Department of Justice filed an application to intervene in Elon Musk's appeal against the €120 million fine the European Commission imposed on X under the Digital Services Act. The filing covers two cases before the EU's General Court in Luxembourg — one

Bill Gates says AI is powerful enough to cause a billion deaths

The Guardrails

Bill Gates says AI is powerful enough to cause a billion deaths

Gates has spent the past month arguing that AI is a policy problem, not a product category. This week he attached a number to it and asked Washington for the law. Separately, Goldman Sachs put a revenue figure on the AI buildout that its spending implies. Bill Gates told NBC's "Meet the Press" that AI is "certainly powerful enough to drive events that, you know, cause a billion deaths" — and that the companies building it should not be left to police themselves. The line came with a mechanism,

Jensen Huang says it doesn't matter if kids forget long division

The Everyday

Jensen Huang says it doesn't matter if kids forget long division

Nvidia's chief executive was handed research showing that AI helps students finish homework without learning the material. His answer was that the skills in question were never worth much — and offered his own life as proof. Jensen Huang's first long interview since Nvidia became the load-bearing company of the AI boom is a case for not mourning what the technology erases. In the New York Times's Ezra Klein Show, published on 23 September, Klein put a Chinese study to him: AI tools let students

Microsoft folds coding and always-on agents into one Copilot app

The Arena

Microsoft folds coding and always-on agents into one Copilot app

Microsoft finally shipped the consolidated Copilot it promised in June — and the surprise is that an app sold to knowledge workers now writes and hosts internal apps. Two other stories worth your time: ICLR's second anonymity incident in two cycles, and Berlin's AI street cameras, which are not judging anything yet. Microsoft unveiled the Copilot "super app" it has been teasing since June: one app with three tabs — Home, Code and Autopilot — that merges Copilot Chat with Cowork and rebrands the

Peking University's 591-run RL study finds uniform sampling wins

The Frontier

Peking University's 591-run RL study finds uniform sampling wins

A Peking University group built the comparison platform, ran the experiments, and reported the boring answer: with everything else held constant, the data policy is not what moves the score. The sharper finding is that the benchmark summary you choose can flip which method looks best. Peking University's Data-Centric AI group ran 591 reinforcement-learning experiments to test whether cleverer data policies beat plain uniform sampling, and could not find one that reliably does. The paper, DataFl

See all