As someone straddling web dev and AI work at a university, this question landed close to home: when an AI agent causes real damage, who actually answers for it?
This piece walks through recent cases of AI agents going rogue in cyberattacks, and the liability picture is messier than I expected. It's not simply "blame the AI vendor." Contracts, terms of service, how much autonomy the agent was given, and whether a human signed off on its actions all shape who ends up responsible.
What stuck with me is that we're deploying agentic tools faster than the legal and organizational frameworks around them are maturing. If I give an agent access to systems here on campus, I need to think past "will it work" and into "who owns the outcome if it doesn't."
Worth a read for anyone rolling out AI agents in their own work, not just the legal team.
Article: https://www.technologyreview.com/2026/09/28/1145197/whos-liable-when-ai-agents-go-rogue/ #AI #Cybersecurity