Skip to main content
Home/Framework/Governing Principles
Part 1 · Foundation

The six Governing Principles

Every requirement in AIGMS traces back to one of six principles. They explain why the framework asks for what it asks for, so they give a business owner a way to judge whether any AI governance arrangement is doing its job.

The six Governing Principles are the test AIGMS applies to its own requirements. Where a rule in the framework would breach one of them, the rule is wrong. They are also the fastest way for a business owner to judge whether an AI governance arrangement is doing its job, whether or not that arrangement is AIGMS.

Principle 01

Proportionality

Governance effort matches what could go wrong

A tool that summarises internal meeting notes and a tool that ranks job applicants do not deserve the same scrutiny. AIGMS™ scales the requirement to the consequence, so a Negligible Activity carries only the six Foundation Safeguards while a Critical one carries the full applicable suite.

What breaches it: Applying identical controls to everything, which usually means either the low-risk work is smothered or the high-risk work is under-governed.
Principle 02

Visibility

You cannot govern what you have not identified

Discovery comes before policy. The Activity Register is built through six channels rather than a single email asking people what they use, because the Activities nobody declares are the ones carrying the unassessed consequences.

What breaches it: A policy that describes how AI should be used, with no record of how it is actually being used.
Principle 03

Human Accountability

A named person owns every Activity

Every entry on the register carries an Accountable Person by name, not a department or a job family. Technology does not absorb responsibility, so an AI-assisted decision remains a decision somebody made.

What breaches it: Ownership recorded as a team, a function or the Governance Lead by default for everything.
Principle 04

Transparency of Purpose

People affected can find out that AI is in use

Where AI contributes to something that affects a person, that person should be able to establish that it did and why. Transparency and disclosure form their own Safeguard Group rather than being left to a privacy notice.

What breaches it: Automation that is deliberately concealed. Disclosure so buried that nobody could reasonably find it counts too.
Principle 05

Continuous Vigilance

Governance is a cycle, not a project

Providers change their models, organisations change their processes and the regulatory position moves. Review frequency follows the consequence rating rather than the calendar, while six defined triggers force a re-profile between scheduled reviews.

What breaches it: Treating implementation as finished, with a register that has not been touched since the month it was built.
Principle 06

Integration

AI governance sits inside the systems you already run

AIGMS™ is designed to run alongside health and safety, data protection and quality arrangements rather than as a separate discipline with its own committee. The Governance Lead is normally an existing manager.

What breaches it: A parallel structure with its own meetings and paperwork, which competes for attention and is usually the first thing dropped.

Using the principles as a check

The principles are most useful when something in the register looks awkward. If an Activity feels over-governed, test it against Proportionality. If nobody can say who owns a tool, that is Human Accountability. If the register has not moved in a year, that is Continuous Vigilance.

A short test. Pick one AI tool your organisation uses. Can you name the person accountable for it, say what could go wrong and point to a record showing what protections are in place? Three yes answers means the principles are being met for that Activity. Anything less shows where to start.

Continue reading

Framework version 3.1  ·  Page last reviewed 28 July 2026  ·  Next scheduled review 28 July 2027
Material changes since the previous version See the framework overview

Put this into practice

The AIGMS packs contain the documents, registers, workbooks and training that turn the framework into a working system.