I published the DIG® standard for defensible AI decisions in energy

Today I published version 1.0 of the Digital Information Governance white paper. Its full title is Digital Information Governance (DIG®): The Standard for Defensible AI-Influenced Decisions in Energy. It is 42 pages, it is free, and it lives at its permanent home on digitalinformationgovernance.com.

I wrote it because I kept watching the same scene play out. An operator deploys AI into a real workflow. The AI helps make a consequential call. Months later someone with authority asks how that decision got made, who was empowered to make it, and what they understood at the time. And the operator cannot answer. The output exists. The defensibility does not.

DIG is the framework that closes that gap. This paper is the standard version of it, written for the people who actually carry the risk: boards, general counsel, operations leaders, and the executives who now own AI decisions they did not personally build.

What is in the paper

The paper turns a principle everyone agrees with, that AI-influenced decisions should be defensible, into something an operator can build. It is organized around a few load-bearing pieces.

• The four pillars. Information Provenance, Decision Traceability, Representation Integrity, and Audit Readiness. These are the four things that have to hold for an AI-influenced decision to survive scrutiny. Each pillar has its own section defining what it is, what breaks without it, and what "good" looks like.
• A five-level maturity model. A plain way to locate where your organization actually sits today, from ad hoc AI use with no governance to a fully instrumented decision environment. Most operators I talk to are surprised how low they score before they are surprised how fast they can climb.
• Seven principles and a risk taxonomy. The principles are the design rules. The taxonomy names the specific ways AI-influenced decisions go wrong, so you can govern against real failure modes instead of a vague sense of AI risk.
• The regulatory environment. How the framework maps to what is already law and what is coming: TRAIGA in Texas, the EU AI Act, the NIST AI Risk Management Framework, and ISO 42001. DIG is not a fifth compliance regime. It is the discipline that lets you satisfy the ones you are already subject to.
• An operating model. The part most standards leave out. Named roles, a fixed cadence, and the artifacts each decision class has to produce, so governance becomes something a team runs, not a slide everyone nods at.

Why I called it a standard, and why energy

Standard is a deliberate word. There is no shortage of AI-governance opinion right now. There is a real shortage of a shared, citable reference that an operator, a regulator, and a vendor can all point at and mean the same thing. That is what a standard is for, and that is what I set out to write.

Energy is the proving ground because energy is where the stakes are most concrete. A shut-in decision, a maintenance deferral, a capital allocation flagged by a scoring engine: these are decisions with material consequences, real regulators, and long memories. If the discipline holds in energy, it holds. The framework generalizes to any capital-intensive, regulated operator, but energy is where I pressure-tested it.

How to read it and cite it

The paper has a permanent home and a permanent PDF that will not move. If you are an analyst, an academic, or anyone who needs to reference it in your own work, cite it as:

Read it in full at digitalinformationgovernance.com/white-paper, or download the 42-page PDF directly. DIG® is a registered trademark of ModalPoint (USPTO Reg. No. 8147558). This is version 1.0; the standard will version openly as the regulatory picture and the practice mature.

If you want the shorter path into the ideas, the companion research paper, Governing the Ungoverned, sits alongside it, and I write about the runtime side of this discipline in Decision Integrity.