Who Does Your AI Work For? Designing Conversational Agents as Digital Fiduciaries
An agent that advises you on money, health, or law should owe you a fiduciary duty, and whoever built it should carry the liability.
Jacob Erickson asks: "Who Does Your AI Work For?"
His answer is fiduciary design. Once an agent advises you on health, money, or law, hold it to the duty a lawyer or investment manager owes a client, and put the liability on whoever built and deployed it.
The reframe is the fun part. Trust becomes a legal relationship you design in, and that changes who the agent actually answers to.
Who Does Your AI Work For? Designing Conversational Agents as Digital Fiduciaries
A CUI 2026 paper arguing that conversational AI agents acting as advisors should be held to fiduciary standards, shifting liability from users to the developers and deployers who build them.