KnowGence®Governing AI Delegation

For Boards

Is your AI structurally governable?

Most boards are told their AI is governed because policies and committees exist. KnowGence shows, with evidence, whether people can still set limits, intervene, roll back and hold someone accountable when AI acts at machine speed.

Why boards are asking

AI now executes decisions faster than traditional oversight can respond. Regulators, auditors and investors increasingly expect boards to show that oversight is real, not just documented. A policy can exist while the control behind it has never been tested.

Four layers, from the boardroom down

01BoardWho holds authority over AI, and is oversight real or only documented?
02OrganizationDoes the enterprise governance architecture hold across all six primitives?
03Each AI systemFor every material model, agent and copilot: an owner, its authority, a finding on each of the six primitives, and the limits it has been shown to stay within.
04Deployed controlsDo the controls running in production, on any platform, do what the board believes they do?

Engagements

2 weeksBoard BriefingAn indicative BAL Score™ and interviews with four to six executives, giving a mandate summary, a named accountability map, the main evidence gaps and priorities for testing. One session with the board or committee and a short findings memo.
6 weeksInstitutional AssessmentThe six primitives tested across material AI systems. Each required control is found Pass, Fail or Unverified on evidence, with no averaging away of a failure; the report shows the authority each system has demonstrated it can be held to, and a remediation roadmap.
12 months, renewableAnnual monitoringRe-testing through the year, with alerts to the board when a model, its authority or the evidence changes, and an annual report that explains any change in scope or method.

Five questions to ask management

  1. Which AI systems can act without a human approving each decision?
  2. How long does it take us to stop one, and when did we last test it?
  3. Who, by name, is accountable for each system’s decisions?
  4. When was each system’s authority last formally re-approved?
  5. What evidence, not policy, shows our controls work?

If any of these are hard to answer, a Board Briefing is the place to start.

The research behind it

Cover of Bounded Agency: Who Is in Control in the Age of AI?

Bounded Agency: Who Is in Control in the Age of AI?

Ravi S. Bhagavatula, PhD, CFA

The research behind every KnowGence assessment: how authority migrates into automated systems, and what boards must do to keep AI in a state of bounded agency, acting within limits people set and can enforce, so that human agency is preserved.

Buy on Amazon (Kindle) →Reader Depot →

Independent by design

KnowGence sells no technology, ranks no vendors and is paid nothing based on which platform a client chooses. Any platform relationships are disclosed.

Our independence commitments →
Request a Board Briefing

Two weeks to a clear view of where your organization stands and what the board should examine next.

Request a BriefingTake the free BAL Score™