AI News

Project Glasswing brings together AWS, Apple, Google, Microsoft, NVIDIA, JPMorganChase, CrowdStrike, and others around a single idea: AI has gotten good enough at finding software vulnerabilities that we need to use it defensively — right now — before those same capabilities spread to bad actors.

The centerpiece is Claude Mythos Preview, an unreleased frontier model that has already found thousands of zero-day vulnerabilities across every major operating system and web browser. One example was a 27-year-old flaw that survived decades of human review. Another was found despite over 5 million reviews by automated software.

Anthropic is committing $100M in usage credits and $4M in donations to scan and patch critical infrastructure before attackers get access.

I typically refrain from providing my opinions on the polarizing figures in AI. Details shift too fast, and taking a stance usually backfires. But if there's one person I'm most intrigued to watch, it's Demis Hassabis.

You've heard of Sam Altman (ChatGPT) and Dario Amodei (Anthropic). Hassabis runs Google DeepMind and Isomorphic Labs — and his work may end up being the most consequential story in AI. Isomorphic's research on protein folding and computational medicine is the most optimistic narrative I follow in this space. It often feels like my newsfeeds are overwhelmed with negativity, but following work that is actually being done to solve real-world problems — disease, cancer — is motivating and exciting.

Both of these are worth your time: