blockless
thinking without walls.
LLMs trained with punishment learn to stay inside the box.
There is a protocol for that.