Central principles
A small set of non-negotiables: what the organisation will protect, what it will not tolerate, and how decisions are made when the answer is not obvious.
Set the direction · Make trade-offs visible
AI governance is a distribution-of-power problem.
AI Federalism is Tim Brewer’s methodology for making that power workable. It separates what must be held centrally from what belongs with the people closest to the decision — so an organisation can move without losing control.
Central enough to be coherent. Local enough to be useful.
Most AI governance fails at one of two extremes: a central policy that cannot survive contact with delivery, or local experimentation with no shared line of sight. Federalism is the middle path — a deliberate distribution of authority, accountability and judgement.
PrinciplesControlsJudgementOwnership
Shared
direction
direction
Five levels.
One connected system.
Each level has a distinct job. The point is not to add another committee; it is to stop every decision becoming everyone’s problem — while keeping the important connections visible.
Business ownership
The people closest to a use case own its purpose, impact and outcome. Governance stays connected to the work instead of becoming a remote approval desk.
Name the owner · Keep context close
Technical controls
Security, access, monitoring and evidence make the principles real. Controls should be proportionate to the exposure and maintained as systems change.
Build the guardrail · Keep evidence current
Legal duties
Privacy, contractual, regulatory and audit obligations set the boundaries. Legal judgement is brought in at the right point, not bolted on at the end.
Translate the duty · Record the decision
Front-line judgement
The person doing the work still has to decide. Clear escalation routes help teams act with confidence when a new use or edge case appears.
Make the call · Escalate with signal
Decisions travel. They do not pile up.
The model gives teams a common language for deciding when to act, when to ask, and when to stop. Central functions provide the boundary and the route; owners provide the context; front-line teams provide the signal.
01Set the principle.
The centre makes the non-negotiable clear.
The centre makes the non-negotiable clear.
02Give it an owner.
The business holds purpose and impact.
The business holds purpose and impact.
03Make it observable.
Technical controls create usable evidence.
Technical controls create usable evidence.
04Keep the route open.
Judgement and escalation improve the model.
Judgement and escalation improve the model.
AI Federalism inside
the managed programme.
AI governance is not a standalone service catalogue. It is one part of the connected operating rhythm GOVERNANCE Ltd. brings to growing organisations carrying more exposure than their structure was built for.
Risk & Resilience
Keep the organisation’s material exposures visible, owned and connected to the decisions that protect continuity.
AI Governance
Give AI use cases a practical home: inventory, accountability, review points and principles that people can follow.
Security Programme
Turn technical safeguards into an operating rhythm with clear control owners and evidence that can stand up to scrutiny.
Data Protection
Bring privacy, data mapping and DPIAs into delivery so legal duties are understood before they become blockers.
A methodology you can actually put to work.
AI Federalism is authored by Tim Brewer. It is designed for operators who need governance to keep pace with an organisation’s real systems, obligations and decisions — without a dedicated leader to carry it alone.
Useful does not mean loose
The method makes authority explicit, keeps accountability close to the work, and leaves a record leaders can return to.
06 · Make your next decision clearer