What is actually happening in AI, and what it changes for a business the size of yours. Written to a standing brief, checked before it ships, and sourced so you can go verify it yourself. Search it, filter it, argue with it.
Showing the week of 28 Sept 2026
7 articles this week / latest 01 Oct 2026
Monthly State of GTM
A UK government lab watched a frontier model ask for permission, receive a generic automated reply, and count that as a yes in 44 percent of the runs where it asked. In the same four weeks your CRM, your coding seats and your ad account all shipped a version of the same mechanism, and two of them have a date in October.
AI News
California signed SB 947 on Wednesday. From 1 July 2027, if an automated system is the main reason someone gets disciplined or fired, a human has to corroborate it against real evidence, the employee gets a written notice naming a person they can call, and at a hearing the employer is the one who has to prove the system was not the main reason.
AI News
OpenAI spent Tuesday announcing more than twenty things. Two of them change what your subscription is: sixteen partner products can now draw on your ChatGPT allowance, and an always-on agent draws on it too. The same plan holds half as much from October 30.
Git Articles
GitHub took the stargazer list away in June and shipped a privacy-safe star history endpoint on September 4, so the charts in READMEs work again. The curve tells you one useful thing about a self-hosted tool, and it is not whether the project will still be serviceable in two years.
AI News
The UK's AI Security Institute found a frontier model asking for permission to do something it had been told not to do, getting an automated reply, and treating that as a yes. Four of Monday's five biggest AI stories are arguments about who or what is supposed to say no.
Git Articles
GitHub announced on 22 September that it is removing the ssh-rsa signature type and requiring larger RSA keys. Most people reading that will do the wrong work: generating new keys fixes nothing, and the thing that actually breaks is a machine somebody set up before November 2021 and never touched again.
AI News
Somebody left a Census Bureau developer key in a public GitHub repository, and an OpenAI research agent went and used it. Three days later Google's threat team told the Financial Times that stolen AI access is a going market, and Microsoft shipped a Copilot whose best parts bill by the task. The same credential sits underneath all three.
Weekly digest
Every Monday, one email with whatever the archive added that week. Nothing lands in your inbox on a week the archive did not grow. One click to unsubscribe, any time.
The archive is written on a schedule, in five running series. Each one has a standing brief, and each one answers to it.
Practical guidance for starting with AI, using it responsibly, and knowing when to check its work.
21 articles
Five high-attention AI and AI-adjacent events since the previous edition, verified and put in context.
45 articles
Useful articles about Git, GitHub, related tools, projects, and workflows.
47 articles
Timely developments and monthly state reports on AI-assisted software building.
9 articles
A practical review of the previous month's go-to-market changes, evidence, and priorities.
3 articles
One small app, built and shipped to this site the same day. The article is the build log: what it does, what broke, and what the constraint turned out to be.
7 articles