// 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.
A new model with the codename Ox Alpha has appeared and parts of the internet immediately dived into guessing who is behind it. Such “stealth” models typically test performance anonymously before an official reveal. For now it is unclear which lab released it.
According to the FT (numbers gathered by Simon Willison), Anthropic struggles to attract users to its best but expensive model, while cheaper tools thrive. Even so, annualized revenue in July rose to about 65 billion dollars from 47 billion in May, and the company expects a profitable third quarter.
TechCrunch unpacks the tangled question of whether training AI models on copyrighted books is legal. Most published authors have, without their knowledge or consent, contributed to the very tools that could undermine their livelihoods. There is no simple answer — legally it is genuinely complicated.
Linus Torvalds described a brutal debug session where AI did much of the tedious grunt-work for him. The funny part: the model repeatedly declared the problem impossible and suggested just writing a report — Torvalds was clearly more stubborn. A neat illustration of where AI helps today and where it still backs off.
Simon Willison points out that the key skill with coding agents is being able to confidently instruct them on a change and then verify it was done correctly. That does not always mean reading every line — eyeballing line by line was never the most effective way. A practical take for teams putting AI into development.
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.