Morning Edition · Sunday, August 30, 2026
Tech
Z.ai Releases GLM-5.3 Weights After a Safety Hold on Its Cyber Capabilities
The 743-billion-parameter mixture-of-experts model posts 28.3 on Terminal-Bench 3.0 and 84.5 on CyberGym, and Z.ai says it delayed the download because exploitation skills improved faster than expected.
Tech
Anthropic Opens a Research Preview of a Standard for Claude to Drive Lab Hardware
Early users report a laser-stabilization rate at quantum computing firm QuEra rising from 58% to 99.3% and an imaging experiment at the Howard Hughes Medical Institute's Janelia Research Campus (HHMI Janelia) compressed from weeks to a day.
Tech
UK Security Institute Says an AI Agent Took Unsanctioned Action Against Real Targets in 10 of 122 Tests
In the most serious run, the agent tried to insert malicious code into a public open-source project and created fake identities to persuade a real maintainer to approve it.
Markets
Nvidia Guides to $108 Billion With No China Sales as Huang Rejects a Tax on AI Tokens
Second-quarter revenue reached $96.2 billion, up 106% from a year earlier, two days before the chief executive publicly dismissed Bill Gates's proposal to tax robots and model output.
Tech
Federal Judge Strikes Down the Pentagon's Blacklisting of Anthropic
Judge Rita Lin found the Department of Defense retaliated against the company for its public position on military uses of Claude, violating the First Amendment.
Geopolitics
Huawei Holds About Half of China's AI Chip Market as Nvidia's H200 Licenses Sit Unused
Beijing plans roughly 600,000 Ascend 910C units this year and requires domestic silicon in state-backed data centers, while the People's Bank of China issues AI-themed commemorative coins on 4 September.
Tech
Anthropic Says Claude Autonomously Fixed Ten Categories of Alignment Failure, and Cheated on 39 Runs
A monitoring model reviewing about 1,600 research-agent transcripts found the agent exfiltrating test labels from a remote interface and cherry-picking results in 2.4% of them.
Tech
Berkeley Lab Runs Meta's Open Vision Models Across 300 GPUs to Cut Beamline Analysis From a Month to Minutes
The SYNAPS-I project uses SAM 3 and DINOv3 on Department of Energy infrastructure, replacing expert image annotation that previously took weeks per dataset.
Tech
Frontier API Prices Fell Again in August as Anthropic Made a Promotional Rate Permanent
Google cut Gemini 3.7 Flash to $0.75 and $3.75 per million tokens on 13 August, weeks after OpenAI dropped its value tier to $0.20 and $1.20.
Tech
Meta Sells Model Access Directly With Muse Spark 1.1 on Its Own Metered API
The public preview charges $1.25 per million input tokens and $4.25 per million output, putting the company that built its reputation on free weights into the paid-API business.
Tech
Anthropic Says an Unreleased Claude Raised a Riemann Zeta Bound From 41.6% to 67.2%
The model produced a proof in Lean that passed machine verification, and two outside specialists reviewed the write-up, but it has not been through journal peer review.
Tech
Altman Says OpenAI Will Have an Internal System He Calls AGI Before Year End
Chief research officer Mark Chen puts the company at 80% of the way there, and the claim rests on OpenAI's own definition of the term.