Sam Altman is the last person who should be writing AI regulation and he knows it.

Sam called for a "new social contract" between AI companies and society, while his company ships gpt-5.4, o3, and operator-class agents at a pace no legislature on earth can track.
aws can't decentralize without cannibalizing its own margins, and every "self-regulatory framework" in tech history has ended with the same few entities writing the rules they agreed to follow
if you're running your model on aws or azure, you're ONE policy memo away from a TOS change that reprices, restricts, or kills your workload
the only architecture that doesn't have that failure mode is permissionless. no single entity decides what can run or which agents get access when trillions of them need compute across billions of devices
a feature announcement wrapped in a warning
Sam's writing the social contract for a stack he owns
builders who don't want to live under it have exactly one option