The DRI Model: Assigning Ownership to AI Systems

Ambiguous ownership is the single biggest operational risk in AI governance. When something goes wrong, if nobody is clearly accountable, response is slow, responsibility is disputed, and nothing gets fixed. The DRI model provides a clean framework for eliminating this ambiguity.

What is DRI?

DRI stands for Directly Responsible Individual. It is a concept borrowed from operational management: for every significant system or decision, one named person is directly responsible. Not a team. Not a department. One person.

• ◆The DRI is not the only person involved — others contribute, review, and support. But the DRI is the person who answers when something goes wrong.
• ◆The DRI has authority commensurate with responsibility — they can make decisions about the system without needing committee approval for routine actions.
• ◆The DRI is named in writing — not assumed based on org chart position. Explicitly documented in the AI Register.

Applying DRI to AI Systems

For each AI system in your register, define two DRIs:

• ◆Business DRI — Accountable for what the system does: the outcomes it produces, the decisions it influences, compliance with regulatory requirements, and responding to harms caused by the system. This is typically a senior business leader.
• ◆Technical DRI — Accountable for how the system works: performance, accuracy, security, monitoring, updates, and technical incident response. This is typically a senior engineer or data scientist.

What Happens Without Clear DRI

• ◆Regulatory investigations have no clear point of contact → response is slow and inconsistent
• ◆Model performance degrades without anyone noticing or acting
• ◆When an incident occurs, time is wasted establishing who is responsible instead of fixing the problem
• ◆Compliance obligations fall through gaps between teams