AI Steering Committee: How to Govern AI Without Creating Another Talking Shop
AI governance often begins with a sensible idea:
Create an AI steering committee.
Bring together representatives from Technology, Risk, Legal, Data, Security, HR, Finance and the business.
Meet regularly.
Review AI activity.
Make decisions.
Coordinate the organisation.
On paper, it sounds straightforward.
In practice, AI steering committees can quickly become something very different.
A monthly meeting.
A long agenda.
Updates from different functions.
Presentations about AI projects.
Lengthy discussions about risk.
Actions that disappear between meetings.
Decisions that nobody is quite authorised to make.
The committee exists.
But AI continues developing around it.
The problem is not necessarily the committee itself.
It is that organisations often create an AI steering committee before defining what decisions the committee exists to make.
A useful AI steering committee should not simply discuss AI.
It should help leadership make better AI decisions and ensure those decisions turn into action.
What is an AI steering committee?
An AI steering committee is a cross-functional leadership group responsible for coordinating and overseeing material artificial intelligence activity across an organisation.
Depending on the organisation, its responsibilities may include:
AI strategy;
portfolio priorities;
investment;
governance;
risk;
ownership;
transformation;
value;
AI agents;
escalation.
The committee should provide a mechanism for bringing together the different functions required to manage AI coherently.
But cross-functional participation alone does not create effective governance.
The committee needs:
a clear mandate;
defined decision rights;
appropriate information;
named owners;
actions;
evidence;
escalation.
Otherwise it risks becoming a forum for AI discussion rather than AI management.
For more on the wider organisational structure required to coordinate AI, read What Is an AI Operating Model? A Practical Guide for Leadership Teams.
Why AI needs cross-functional leadership
AI rarely belongs entirely to one function.
A material AI initiative might involve:
Business — owns the outcome.
Technology — enables the capability.
Data — provides information.
Security — protects systems and access.
Legal — considers legal implications.
Risk — provides challenge and oversight.
HR — manages workforce implications.
Finance — challenges investment and value.
That creates a coordination problem.
If every function makes decisions independently, AI can become fragmented.
If every decision requires everyone, AI becomes slow.
The steering committee therefore needs to balance:
COORDINATION
with
DELEGATION.
Its purpose should not be to centralise every AI decision.
It should establish where decisions belong and ensure material issues reach the right level.
The first question: what is the committee actually for?
Before deciding who attends, define the mandate.
A weak mandate sounds like:
Provide oversight of AI.
That can mean almost anything.
A stronger mandate might be:
Ensure material AI investment and activity aligns with business strategy, has clear ownership, operates within agreed governance boundaries and produces appropriate evidence of value and risk.
That gives the committee something concrete to manage.
The Stratify AI Steering Framework
Structure the steering committee around seven recurring responsibilities:
ALIGN → PRIORITISE → AUTHORISE → OVERSEE → UNBLOCK → EVIDENCE → DECIDE
Together, these answer:
What does leadership actually need to do to keep enterprise AI moving in the right direction?
1. ALIGN — Is AI still connected to strategy?
The committee should continually connect AI activity to business priorities.
Ask:
Which strategic outcomes are we trying to improve?
Are our AI investments supporting them?
Have business priorities changed?
Are we pursuing AI because it matters or because the technology is available?
The committee does not need to recreate the AI strategy every month.
It needs to ensure execution remains connected to it.
For more on why AI needs to remain anchored to business outcomes, read AI Strategy Without Business Value Is Just a Technology Plan.
2. PRIORITISE — Are we working on the right things?
AI opportunities will usually exceed available:
capital;
technology capacity;
data capacity;
change capacity;
leadership attention.
The committee therefore needs visibility across the portfolio.
It should understand:
what is being explored;
what is being piloted;
what is in production;
what is scaling;
what is blocked;
what is consuming investment.
The question is not simply:
Is Project A progressing?
It is:
Is Project A still one of the best uses of our resources?
For more on managing AI opportunities and investments collectively, read How to Build an AI Portfolio: Managing AI Opportunities, Pilots and Investments as One System.
3. AUTHORISE — Which decisions actually belong here?
This is where many steering committees become bottlenecks.
They attempt to approve:
every AI tool;
every experiment;
every use case;
every data request;
every change.
That is not scalable.
Decision authority should be proportionate.
For example:
LOW-RISK ACTIVITY
Proceed within approved boundaries.
MODERATE-RISK ACTIVITY
Functional review or registration.
HIGH-RISK ACTIVITY
Formal approval.
ENTERPRISE / CRITICAL DECISION
Steering committee or executive decision.
The steering committee should therefore spend more time designing delegated authority than exercising authority over every individual case.
For more on determining who should approve, override or stop AI, read AI Decision Rights: Who Should Be Allowed to Approve, Override or Stop AI?
4. OVERSEE — Where does leadership attention need to go?
The committee needs visibility of material:
risk;
investment;
agents;
incidents;
dependencies;
adoption;
value gaps.
But it should not review everything equally.
Leadership attention should follow materiality and exception.
If a low-risk use case is operating within approved boundaries and producing expected outcomes, it may require very little committee time.
If a high-risk AI agent has:
new permissions;
unexpected behaviour;
weak monitoring;
significant customer impact,
leadership attention should increase.
The objective is management by exception, not management by meeting.
For more on establishing the wider governance capabilities required around material AI, read AI Governance Checklist: 15 Questions Every Leadership Team Should Ask.
5. UNBLOCK — What is preventing progress?
A steering committee should not exist only to control AI.
It should help enable it.
Common blockers may include:
data access;
integration;
unclear ownership;
slow approvals;
resource constraints;
conflicting priorities;
missing skills;
procurement;
security reviews.
The committee brings together people with enough authority to resolve cross-functional barriers.
That makes unblocking one of its most important responsibilities.
A useful steering committee should therefore ask every meeting:
What is stuck that requires leadership intervention?
If nothing ever gets unblocked, the committee may be observing the organisation rather than managing it.
6. EVIDENCE — What do we actually know?
AI steering committees can become dominated by opinions.
Project sponsors believe their initiatives are progressing.
Technology believes performance is strong.
Users say productivity has improved.
Leadership needs evidence.
For material initiatives, ask:
What was the baseline?
What was expected?
What adoption has occurred?
What operational outcome changed?
What value has been validated?
What risks or exceptions have emerged?
Evidence should improve as investment increases.
For more on converting AI activity into measurable business outcomes, read AI Benefits Realisation: How to Turn AI Productivity Into Measurable Financial Value.
7. DECIDE — What happens next?
Every material AI initiative should eventually require a decision.
SCALE
Increase investment or deployment.
CHANGE
Alter the technology, process, control or approach.
PAUSE
Wait for evidence or dependencies.
STOP
End further investment.
The steering committee should create a regular mechanism for making these decisions rather than allowing initiatives to continue indefinitely because nobody explicitly stopped them.
For more on making these continuation decisions consistently, read When Should You Stop an AI Project? A Framework for Scale, Change or Stop Decisions.
The Stratify AI Steering Cycle
The framework therefore becomes:
ALIGN
Are we pursuing the right outcomes?
↓
PRIORITISE
Are these the right opportunities?
↓
AUTHORISE
Are decision rights clear?
↓
OVERSEE
Where does leadership attention belong?
↓
UNBLOCK
What is preventing progress?
↓
EVIDENCE
What do we actually know?
↓
DECIDE
What happens next?
Then repeat.
ALIGN → PRIORITISE → AUTHORISE → OVERSEE → UNBLOCK → EVIDENCE → DECIDE
This turns the steering committee from a meeting into a management cycle.
Who should sit on an AI steering committee?
There is no universal membership.
It depends on the organisation and its AI activity.
Core representation might include:
Executive AI sponsor;
Business leadership;
Technology;
Data;
Risk;
Security;
Legal/Compliance;
HR;
Finance;
Transformation.
But not everybody needs to attend every discussion.
A permanent committee of 20 people can become difficult to operate.
A better model may be:
CORE MEMBERS
Own recurring enterprise decisions.
SPECIALISTS
Join when relevant.
BUSINESS OWNERS
Attend when their initiatives require decisions.
The objective is to bring the right authority into the room, not the greatest number of people.
The CEO should not become the default escalation route
If every difficult AI decision reaches the CEO, the operating model is incomplete.
The organisation needs delegated authority.
The steering committee should help define:
what functions can decide;
what business owners can decide;
what specialist functions approve;
what the committee decides;
what genuinely requires CEO or Board attention.
For more on avoiding accountability gaps while distributing responsibility, read The AI Accountability Gap: 7 Signs Nobody Really Owns AI.
The Board and steering committee have different jobs
The Board should not become the enterprise AI steering committee.
The steering committee manages and coordinates AI within the organisation.
The Board provides appropriate oversight of material:
strategy;
investment;
risk;
governance;
performance.
The steering committee should therefore provide the Board with a concise view of the issues requiring Board-level attention rather than forwarding operational detail.
What information should the committee see?
A useful steering pack might contain:
STRATEGY
Are AI priorities still aligned?
PORTFOLIO
What material initiatives exist?
INVESTMENT
What are we spending?
OWNERSHIP
Who owns each outcome?
RISK
What requires attention?
ACTIONS
What is overdue or blocked?
VALUE
What evidence is emerging?
DECISIONS
What needs approval or Scale / Change / Pause / Stop?
The committee should not need six teams to spend a week manually building this view before every meeting.
The AI Steering Decision Log
Every material decision should be captured.
For example:
DATE
When was the decision made?
DECISION
What was decided?
RATIONALE
Why?
OWNER
Who owns the outcome?
ACTION
What happens next?
EVIDENCE
What information supported the decision?
REVIEW
When should the decision be revisited?
This creates organisational memory.
Six months later, leadership should not have to ask:
Why did we approve this?
The evidence and rationale should be visible.
The AI Steering Action Register
Discussion without action is one of the easiest ways steering committees lose value.
Every action should have:
ACTION
What needs doing?
OWNER
Who is accountable?
DUE DATE
When?
PRIORITY
How important?
DEPENDENCY
What is blocking it?
EVIDENCE
What demonstrates completion?
STATUS
Not Started / In Progress / Blocked / Complete.
Don't measure committee effectiveness by meetings held
A steering committee should not be judged by:
attendance;
number of meetings;
number of papers reviewed.
Measure whether it improves AI management.
Possible indicators include:
decision time;
blocked actions;
portfolio decisions;
ownership gaps;
overdue governance actions;
benefits realised;
incidents;
time-to-scale;
initiatives stopped or reallocated.
The committee exists to improve decision quality and execution.
The Fractional CAIO and the steering committee
A Fractional CAIO can help coordinate the steering process without replacing executive ownership.
The role might:
prepare the portfolio view;
surface priorities;
identify accountability gaps;
coordinate actions;
bring evidence together;
identify decisions required;
track follow-through.
The committee makes the appropriate organisational decisions.
The Fractional CAIO helps ensure the management system operates between meetings.
That distinction is important.
A monthly steering meeting cannot manage enterprise AI by itself.
The work between meetings is what turns decisions into outcomes.
10 Questions to Test Your AI Steering Committee
Leadership should ask:
Does the committee have a clearly defined mandate?
Is AI activity connected to enterprise business priorities?
Can the committee see the material AI portfolio in one place?
Are decision rights sufficiently clear that low-risk activity does not require unnecessary escalation?
Does the committee focus attention according to materiality and exception?
Can it resolve cross-functional blockers rather than simply discuss them?
Are material AI decisions supported by evidence rather than opinion alone?
Are actions assigned to named owners with deadlines and evidence of completion?
Does the committee routinely make Scale / Change / Pause / Stop decisions?
Can leadership demonstrate that steering decisions are improving AI outcomes, governance or value?
If several answers are no, the organisation may have an AI steering meeting without having an effective AI steering system.
AI steering should accelerate good decisions
An AI steering committee should not exist to create another layer of bureaucracy.
Its purpose is to help the organisation:
focus;
coordinate;
decide;
unblock;
learn;
adapt.
The best steering committees should make lower-risk AI easier to pursue because boundaries and decision rights are clear.
And they should make higher-consequence AI more visible because leadership knows where attention is required.
The objective is not:
More governance meetings.
It is:
Better AI decisions, made at the right level, with evidence and follow-through.
Frequently Asked Questions About AI Steering Committees
What is an AI steering committee?
An AI steering committee is a cross-functional leadership group responsible for coordinating material AI strategy, investment, governance, risk, transformation and value decisions across an organisation.
Does every organisation need an AI steering committee?
Not necessarily. The appropriate governance structure depends on organisational size, complexity, AI activity and existing leadership arrangements. The important requirement is that cross-functional AI decisions have clear ownership and coordination.
Who should be on an AI steering committee?
Membership may include representatives from Business, Technology, Data, Risk, Security, Legal, HR, Finance and Transformation. The committee should remain small enough to make decisions effectively, with specialists joining where required.
What should an AI steering committee do?
A steering committee can align AI with strategy, prioritise the portfolio, establish decision authority, oversee material issues, remove blockers, review evidence and make continuation or escalation decisions.
How often should an AI steering committee meet?
Frequency depends on the pace and materiality of AI activity. Some organisations may require monthly meetings, while others may use a quarterly cadence supported by more frequent operational governance.
Should an AI steering committee approve every AI use case?
Generally no. Lower-risk AI should be capable of operating within delegated policies and authority. The steering committee should concentrate on material decisions and exceptions.
What should an AI steering committee report to the Board?
Board reporting may include AI strategy, major investments, material risks, significant incidents, governance gaps, performance, realised value and issues requiring Board-level oversight.
What is the difference between an AI steering committee and an AI Centre of Excellence?
A steering committee primarily provides cross-functional leadership, governance and decision-making. A Centre of Excellence typically provides expertise, standards, enablement, technology support or implementation capability.
Who should chair an AI steering committee?
The appropriate chair depends on the organisation. It may be a CAIO, CIO, COO, transformation leader or another executive with sufficient enterprise mandate. The important factor is authority to coordinate across functions.
How do you know whether an AI steering committee is effective?
Look beyond meetings and attendance. Effective steering should improve decision speed, portfolio focus, accountability, action completion, risk visibility and evidence of AI value.
Make Better AI Decisions Across Your Organisation
AI leadership requires more than understanding the technology. It requires the ability to align priorities, make decisions, assign ownership and turn AI activity into measurable outcomes.
Take our free 5-minute Executive AI Leadership Assessment
Discover where your organisation is strong, where gaps exist and what to prioritise next.
Comments