About

What this is

A directory of the people and shows doing the most credible work on AI safety, agent safety, AI security and governance — plus three original pieces a week that connect what they're saying.

How we write

Every article starts from things our directory voices published in the last two weeks. We don't summarise single sources. Each piece must compare at least two voices, add missing context, map a disagreement, or turn expert views into a practical checklist.

Drafts are written with AI assistance against a fixed brand guide, then pass automated checks before publishing: every quote is verified word-for-word against the source transcript, quotes are kept short, copied phrasing is detected and rejected, and every source is linked. Pieces that fail are held for human review.

Corrections

Corrections and source updates are reviewed before the next publication. The contact channel will be added before public release.