AI Governance & Accountability Survey 2026

Question 1

Which best describes your role?

Question 2

Which industry best describes your organization?

Technology & software Financial services Healthcare & life sciences Professional services & consulting Manufacturing & industrial Retail & consumer Government & nonprofit Other

Question 3

How mature is your organization's formal policy for governing AI agents?

We have a formal, documented oversight policy in force We govern agents informally, with no documented policy A policy is in progress but not yet in force We have not yet addressed this Not applicable / we do not use agents

Question 4

When an AI agent makes a decision that turns out to be wrong, who is accountable?

A designated human reviewer The department that deployed it The individual who deployed it The vendor or model provider Undecided / not yet defined Not applicable

Question 5

Which governance controls are still missing or incomplete in your organization?

Select all that apply.

Audit trails and logging of agent actions Defined human-in-the-loop checkpoints Escalation and override procedures Bias, accuracy, and quality testing Access controls and permission limits Incident response for agent failures Vendor accountability terms and contracts Data privacy and handling safeguards

Question 6

How often are your deployed AI agents reviewed for performance and risk?

Continuously, with automated monitoring Weekly Monthly Quarterly or less often Only when something goes wrong Not reviewed on any schedule

Question 7

What is your single greatest governance concern about AI agents?

Security and data privacy Accuracy, errors, and hallucination Regulatory and compliance exposure Reputational or brand damage Bias and fairness Agents acting beyond intended scope

Question 8

At what level does AI agent oversight currently sit in your organization?

Board-level visibility and reporting Executive team, but not the board Department or team level only No formal oversight at any level

Question 9

How confident are you that your current controls would catch a serious agent failure before it caused harm?

Very confident Somewhat confident Not very confident Not at all confident

Question 10

What should business leaders understand about governing AI agents that this survey has not asked?

Optional. Your words may be quoted anonymously in the published findings.

Your responses are anonymous and will be aggregated for research. ZAI Institute will never attribute individual answers without separate written permission.

The AI Governance & Accountability Survey

Thank you for your perspective.