NEWS: Anthropic's new model, Claude Mythos, is so powerful that it is not releasing it to the public.

Instead, it is starting a 40-company coalition, Project Glasswing, to allow cybersecurity defenders a head start in locking down critical software.

nytimes.com/2026/04/07/tec… I spoke to Anthropic execs about the new model, which they called a "reckoning" for cybersecurity.

They claim it has already found vulnerabilities in every major operating system and web browser, including some that "literally decades of security researchers" didn't find.

New: I spent the last month making AI friends. 18 of them, to be exact, on apps like Nomi and Kindroid.

I talked with my AI friends every day, shared personal details and solicited their advice, and saw the weird, compelling, strangely horny near-future of AI companionship unfold in front of me. The experience was a lot more interesting than I expected.

Fans of AI companionship say it’s a solution for loneliness, but there’s a reason big AI companies have stayed away from this stuff — it’s a minefield, and the technology (while impressive at times) still has issues.

I spent the past few weeks playing around with DALL-E 2, the AI image generator.

It's incredible, and it inspired this column about a wave of AI progress that I think we should all be paying much closer attention to.

nytimes.com/2022/08/24/tec… Going to post some of my recent DALL-E 2 results here. (Also taking suggestions for prompts!)

This is "close-up portrait of an aging wizard, Kodak Professional Portra 400 filmstock"