The repeated argument that AI companies are talking about existential risks as a way of creating hype & inflating their stock prices has alw...

Genie 4 runs DOOM: "Walking through a typical office but every computer is running DOOM and all of the walls are also running DOOM and also ...

Genie 4 is quite an insane experience, generating game-like enviornments as you go. It isn't flawless but I can type "Neogothic brutalist ci...

This never gets old (& also shows how hard forecasting AI is even for AI) “Hey Claude Fable & GPT-6 Astra, Without looking anything up, whe...

The end of detailed summarized thinking traces for Claude and ChatGPT (likely both anti-distillation & for UX) has made being a human in the...

Given how out-of-nowhere it seems to many, it is astonishing that there was a small group of people who predicted with remarkable accuracy e...

I had GPT-6 Pro revisit the cool finding that violent video games reduce crime, as people stay home playing them. It downloaded fresh crime ...

My general advice for using AI for any purpose is to ask yourself: "how do I make what I am doing even more ambitious?" You do this until y...

The website for my book redone with Opus 5.5. The initial version (made with Opus 4.8) required a lot of work to make a distinctive-looking ...

A thing I see happening at large organizations as a result of AI is the expansion of the role and span of control of senior management who c...

Without being overly credulous, we should be taking seemingly “out there” ideas about AI cybersecurity from the Labs more seriously. Example...

As the gap between the internal models at the AI labs and the ones they release grow (due to safety concerns & research advantages), it will...

The inventor of the term "prompt injection" suspects (as do I) that this may not be as large a problem in new models. This matches our tests...

By every measure, Opus 4.8 beats human experts as peer reviewers in a study of 124 Harvard social science papers covering a range from stati...

We don't have a lot of evidence, but the bulk of the academic papers modelling the dynamic seem to suggest AI initially helps attackers mor...

Very useful prompt for Codex with computer use.

When my first AI book was published in 2024, the best AI could do 4 minutes of human work with a 50% chance of success When I turned in the...

A visible sign of what AI insiders have been warning about since at least the Mythos release. Acceleration of attack and defense, at inhuman...

This is an interesting question because the answer is not obvious. Which, by the way, does not mean AI coding tools are not leading to mass...

Companies I talk to are struggling with the rate of patches they need to do thanks to AI defenses finding issues before the inevitable wave ...