Arvind Narayanan lists 14 AI safety points from Ezra Klein interview

Princeton computer scientist Arvind Narayanan lists 14 AI safety points in a post about his appearance on the Ezra Klein show.

He says AI power depends on what systems are given access to, not on the model alone, and that more investment in control measures such as sandboxes and monitoring would have improved safety.

Narayanan also says defensive AI agents will be needed, and he is skeptical that superintelligence will achieve superhuman persuasion. He says current policy responses to warnings remain inadequate.