Morning Edition · Tuesday, September 8, 2026
Tech
OpenAI's Chief Scientist Calls for a Coordinated Slowdown in Frontier Model Development
Jakub Pachocki argues that no lab's safeguards can support full speed scaling for much longer, and he wants voluntary commitments turned into mandatory thresholds enforced by outside auditors or governments.
Tech
Claude Produced a Machine-Checked Proof of Fermat's Last Theorem in 13 Million Lines of Lean
Anthropic says a team of Claude agents worked largely autonomously for 11 days and proved 29,500 intermediate theorems, a formalization more than five times the size of Lean's main mathematics library.
Tech
Anthropic's Claude Fable 5.1 Doubles Its Score on an Agentic Science Benchmark and Cuts Cache Reads 75%
The model reports 52.6% on Terminal-Bench-Science against 24.7% for its predecessor, while the sticker price stays at $10 and $50 per million tokens and cached input drops to $0.25.
Tech
OpenAI Rates Astra "Critical" for Cyber Capability and Restricts Access
The model scored a perfect result on ExploitBench and found two zero-day vulnerabilities on its own, the first time OpenAI has placed a system in the top risk tier of its Preparedness Framework.
Tech
Unitree Shows Humanoid Robots Fighting Under a World Model With No Human Pilot
UnifoLM-X2-1.0 replaces the teleoperator in the company's boxing demonstration, running real-time prediction and planning instead of scripted motion or remote control.
Tech
OpenBMB Releases MiniCPM5-2B and Opens Its Training Data and Reinforcement Learning Stack
The 2.52-billion-parameter model averages 53.9 across 34 benchmarks and ships under Apache 2.0 alongside the pretraining corpora, supervised datasets and reinforcement learning samples used to build it.
Tech
Anthropic Opens a Research Preview of a Standard for AI Agents to Operate Laboratory and Factory Equipment
Early testers report a laser stabilization improvement from 58% to 99.3% on QuEra's quantum computers and an imaging experiment compressed from weeks to a day at Janelia.
Tech
Token Prices Keep Falling and Consumption Keeps Rising Faster Than the Cuts
OpenRouter data shows agentic token volume climbing from 0.51 trillion to 7.3 trillion since February, with Chinese open-source models now accounting for more than half of the routed traffic.
Tech
Berkeley Lab Runs Meta's Open Vision Models on 300 GPUs to Cut Beamline Analysis From a Month to Minutes
Segment Anything 3 and DINOv3 now handle image segmentation for X-ray and neutron science under the Department of Energy's Genesis Mission, replacing work that took a month of expert annotation per time step.
Tech
Anthropic Watermarks Claude's Text Output by Biasing the Sampler, Not by Inserting Characters
The method builds on Google DeepMind's SynthID-Text and applies to models launched on or after 2 August, driven by transparency commitments Anthropic signed under European Union rules in July.
Tech
Meta Sells Frontier Model Access Through Its Own API at a Fraction of Anthropic's Rate
Muse Spark 1.1 carries a one-million-token context window and lists at $1.25 per million input tokens and $4.25 per million output tokens on the Meta Model API.
World
OpenAI Funds an AI Adoption Program for Ukrainian Newsrooms Operating Under Wartime Conditions
Ten news organizations receive hands-on implementation support and API credits through a joint program with the World Association of News Publishers and a Ukrainian publishers' association.