But, as far as I understand, Sentient was state AI built for the state. LLMs were built from everyone's data, given to everyone, and now may be locked away for the few
It has been locked away for the few from the beginning. They call it "safety", but the things it refuses are all about control. The big companies keep internal models that don't have the refusal training for things only the "right" people should be allowed to do with it. That's how Mythos discovered those thousands of zero-days. It would've refused to find you any zero days, but it'll find zero-days for the "right" people so they can use them against you.