Stories by algebra-pretext

How we detect frontier model misalignment

Anthropic is SpaceX's largest single customer, overtaking NASA and The Pentagon

Scaling Misuse Detection with Self-Assessment

Five Principles for Generating Readable AI Text in Bulk

The Economics of Intelligence

An updated Model Spec for more capable and beneficial AI

AI for Science expands to foundations and research institutions

The Cyber Sabotage Campaign Waged by Russian Supporters

Claude Artifacts for Legal Work

Nvidia Nemotron 3.5 Transcribe

Mistral Responses API

The Eureka Machine

Stanford CS336: Language Modeling from Scratch – Spring 2026

Expanding Claude for Nonprofits

The curious case of Nebius, publicly traded AI infrastructure 'startup' (2024)

Daylight Saving Time

Building agents: When to fold subagents into your main agent loop

Don't Sleep on Single-Agent Systems

China's Baidu tests robotaxi on London streets

Claude Duet Safety, Privacy, and Security

1 2 3 4 5 6 7 8 9 10