Governance Research
How AI oversight works in practice, not on paper
Most organisations now have an AI policy. Far fewer can show a decision that policy changed. Our governance research is concerned with the difference.
Research questions
What we are studying
Does the gate hold?
Whether documented approval gates actually stop deployments, or are routinely waived under delivery pressure.
Who is accountable?
Whether the named owner of an AI system has the authority to halt it, and whether that authority has ever been exercised.
What does the board see?
How AI risk is reported upward, what is filtered on the way, and whether escalation thresholds have ever been met and reported.
Where is the shadow AI?
The gap between the AI inventory an organisation maintains and the AI it actually depends on, including AI embedded in purchased software.
Does assurance change anything?
Whether independent findings result in changed systems, or in documented acceptance of the same risk.
What survives a year?
Which governance controls are still operating twelve months after implementation, and which quietly lapsed.
The failure mode is almost never the absence of a policy
It is a policy with no owner, no gate, no evidence, and no consequence for bypassing it. Governance research that only counts policies will not find this.
Related
Deploy AI with confidence.
Start with an independent assessment of your highest-stakes AI system.