Simon Willison: AI agents cannot be 'Directly Responsible Individuals' — accountability requires a human subject

Main idea

Simon Willison argues that the DRI (Directly Responsible Individual) management framework — the person ultimately accountable for a project's success or failure — cannot be applied to AI agents. Accountability is conceptually bound to a human subject; a machine cannot bear responsibility for its decisions.

Context

The post appeared July 12, 2026, amid growing pressure to delegate decision-making to AI agents in enterprise settings. Willison invokes IBM's 1979 internal training slide: 'A computer can never be held accountable, therefore a computer must never make a management decision.' This quote resonates as a key principle of caution in the age of agentic AI, where accountability can dissolve across agent chains.

Why it matters

For developers and IT managers building agentic systems, this is a critical warning: without a clearly defined human DRI for each agentic task, organizational risk emerges — no one is accountable when an agent makes a mistake. Essential for designing AI agent governance frameworks.

Details / arguments

• DRI framework originates from Apple (Steve Jobs era) — one person clearly responsible for the outcome
• IBM 1979: 'A computer can never be held accountable, therefore a computer must never make a management decision'
• An AI agent can execute a decision but cannot bear consequences — accountability falls on the human, not the system
• Takeaway: when deploying AI agents, always define a human DRI for each agentic workflow