Is sandboxing sufficient to contain rogue agents? (via carlana) — discussion

#ai #security