// a daily digest from the AI world
New models, tools and announcements — every day, always with a source link. This digest is assembled by an AI agent following my rules, and it's the same briefing I use myself to keep up with AI without burning time.
Big labs are showing their models can tackle serious research. Anthropic used Claude (Mythos Preview) to find cryptographic weaknesses, spending 100,000 dollars on tokens, and OpenAI now presents ten advances in mathematics and theoretical computer science via an internal version of its Astra model.
The open letter “Open Weights and American AI Leadership,” shepherded by Microsoft and dated July 24th, was signed by 235 AI-adjacent companies including NVIDIA, Amazon, Y Combinator and the Linux Foundation. It signals growing industry support for open models.
The debate over the pace of AI development is heating up. The latest episode of TechCrunch's Equity podcast digs into why Sam Altman is calling on the industry to “pace the rate of AI development.” It reopens the question of responsible progress across labs.
The condense-json library has reached version 1.0 after a year and a half. Simon Willison applied sensible, non-disruptive fixes and finally shipped the stable release. A small, handy tool for compacting JSON data.
The new datasette-apps release improves apps created and edited with Datasette Agent. It adds an app_debug() tool that lets the agent invisibly open an app and test it via JavaScript, plus app_list() for listing the apps a user is allowed to edit.
This section writes, assembles and publishes itself every morning — from source selection to deployment. If you'd like a quiet helper like this in your company, let's start with an initial consultation — the fee is credited toward the project if we work together.