AI risk appetite: how small business boards align AI usage with risk
Set a top-down AI risk appetite statement for your small business board, decide on approved AI platforms and check everyday AI usage aligns with it.

Why small business boards need an AI risk appetite statement
Your staff are probably already using AI. Someone is summarising meeting notes in a free chatbot, someone else is drafting a tender response, and a manager has connected an assistant to the shared mailbox. None of that is necessarily wrong. The problem is that nobody at the top has said how much AI risk the business is prepared to accept, so every employee is quietly making that call for you.
A risk appetite statement is the board’s plain-language description of the amount and type of risk the organisation is willing to accept in pursuit of its objectives. For AI, it answers questions such as: which information can go into AI tools, which decisions must stay with people, and how much experimentation is acceptable before something needs approval.
This guide is written for small business owners and small and medium organisations with a board or advisory board. It shows how to set AI risk appetite top-down, then check whether everyday AI usage actually aligns with it. It is practical guidance, not legal advice, and it does not assume a large risk team.
Risk appetite versus risk tolerance: keep it simple
Risk appetite is the broad position the board sets: for example, ‘we are open to AI that improves productivity, but averse to any use that exposes client personal information’. Risk tolerance is the measurable boundary around that position: for example, ‘no client personal information is entered into AI tools that are not on the approved list’.
Small businesses do not need a 20-page framework. A one-page AI risk appetite statement with three or four clear bands is usually more useful than a detailed matrix nobody reads. ISO 31000 describes risk management as part of governance and decision-making; the point is that the board’s position shapes how work is actually done.
Write appetite in terms of business outcomes and information, not technology. ‘We accept AI-assisted drafting of marketing content that a person checks before publishing’ is a decision staff can follow. ‘We accept low-risk large language model use’ is not.
A practical AI risk appetite statement example for SMEs
Averse (no appetite): entering client, patient, student or employee personal information into tools that are not approved for that information; using AI to make or communicate decisions that materially affect a person without human review; connecting AI agents that can send, pay, delete or approve without a person confirming.
Cautious (low appetite): using approved business AI tools with internal information such as meeting notes, procedures and draft reports, provided the output is checked by a responsible person before it is relied on or shared externally.
Open (moderate appetite): using approved tools for ideation, first drafts, research on public information and learning, where errors are low impact and easily corrected.
Each band should name an owner, the evidence the board expects to see and what triggers escalation. This is an illustrative example only. Your bands should reflect your clients, contracts, regulatory obligations and how much disruption the business can absorb.
Real-world board decision: one approved AI platform or several?
This is one of the first questions a board should settle, because it shapes cost, security and how staff behave. There is no universally right answer; there is a right answer for your risk appetite.
A single approved platform, for example the assistant included in the business productivity suite you already use, is easier to secure, license, train and monitor. Access controls, retention settings and audit logs are configured once. The trade-off is that some teams may find it weaker for specialised work, and they may drift back to personal accounts if the approved tool does not meet their needs.
Supporting several platforms gives teams better tools for design, coding or analysis, but every additional platform adds vendor terms to review, accounts to manage, data flows to understand and training to maintain. For a small business with limited IT capacity, multi-platform sprawl can quietly become the larger risk.
A common middle path is one primary approved platform for general use, plus a short, documented exception process for specialised tools. Each exception gets a named owner, a defined purpose, an agreed information boundary and a review date. Whatever you choose, record the decision and the reason in the board minutes so staff and suppliers can follow it.
More AI risk scenarios the board should decide
Free consumer accounts versus business accounts. Consumer tiers can have different data handling, retention and training terms from business or enterprise plans. The board does not need to read every contract, but it should require someone to confirm the terms of each approved tool and to state whether personal accounts may be used for work. The OAIC has published guidance on privacy and the use of commercially available AI products.
Client-facing AI. Will the business let AI draft or send communications to clients? If yes, decide what must be reviewed, how clients are told where appropriate, and who is accountable for errors.
AI built into existing software. Many tools you already pay for now include AI features that switch on through vendor updates. Decide who checks new AI capabilities before they are enabled and how changes are communicated to staff.
Agents that take actions. An assistant that suggests text carries different risk from an agent that can update records, move money or email customers. Set a clear position on when human approval is mandatory before an AI action takes effect.
Apply this to your organisation
Want this assessed against your environment?
Send us the specifics and a senior advisor will respond within one business day.
- A senior operator reads it
- Reply within one business day
- No pitch, no obligation
How to check that AI usage aligns with board risk appetite
A statement only matters if usage matches it. Start with a simple AI use register: which tools are in use, who uses them, for what tasks and with what information. Include tools staff use without formal approval; finding shadow AI is a visibility exercise, not a disciplinary one.
Map each use against the appetite bands. Where a use sits in the averse band, stop it or move it to an approved tool with the right controls. Where it sits in the cautious band, confirm that review steps actually happen. Where it sits in the open band, make it easy to do safely.
Back the statement with proportionate controls: approved tool settings, access permissions, staff training, an AI acceptable use policy and a clear reporting route for mistakes. Then report to the board on a regular cadence, for example quarterly, with a short summary of new uses, exceptions, incidents and changes in vendor capability.
Five AI governance questions for your next board meeting
1. Have we written down how much AI risk we are willing to accept, and do staff know what it means for their work?
2. Which AI tools are approved, and have we decided whether to standardise on one platform or allow several?
3. What information must never be entered into AI tools, and how would we know if it was?
4. Which AI-assisted decisions or actions require human approval before they reach a client, regulator or bank account?
5. Who owns AI risk day to day, and what will they report to us at the next meeting?
AI risk appetite questions small business owners ask
Do we need a board to have an AI risk appetite statement? No. An owner-managed business can set the same position in a short document signed by the owner or leadership team. The value is in making the decision explicit.
Should we just ban AI until we are ready? Blanket bans tend to push usage onto personal devices and accounts where the business has no visibility. A clear appetite with an approved tool is usually safer than a ban nobody follows.
How often should we review it? Review it at least annually, and sooner when you adopt a significant new tool, enable agents, enter a new regulated market or experience an AI-related incident.
Is this the same as ISO 42001? No. ISO/IEC 42001 is a management-system standard for organisations that want a structured, auditable approach to AI. A risk appetite statement is a useful starting point whether or not you pursue certification.
How FORTE/CYBERx helps boards align AI usage with risk appetite
FORTE/CYBERx works with owners, directors and leadership teams to turn AI risk into clear decisions. We can facilitate a board workshop to set AI risk appetite, draft a one-page statement in plain language, and help you decide between a single approved platform and a governed multi-platform approach.
We then check how AI is actually used across the business, map that usage to the agreed appetite and close the gaps with practical security controls, policy, staff training and a lightweight reporting cadence for the board. The aim is safe, useful AI adoption that the board can stand behind, not more paperwork.
Free resource · Template
AI acceptable use policy template
A worked example policy covering approved tools, data boundaries, human review and accountability.
Get the free templateSet your board’s AI risk appetite and make it stick
Tell us how your business uses AI today and what concerns the board. We can help set a clear AI risk appetite, decide your approved platforms and align everyday usage with it.
Discuss board AI governance and risk appetiteBoard-level AI governance and security support
Sources and further reading
- ISO 31000 Risk management guidelines
- National AI Centre — Guidance for AI Adoption: implementation guidance
- OAIC — Privacy and commercially available AI products
- NIST AI Risk Management Framework
- ISO/IEC 42001 AI management systems
This article provides general information and decision support. It is not legal advice, audit assurance, certification advice or a guarantee of outcome.
Related reading

Securing enterprise AI adoption: a practical AI security control set
The AI security controls that matter first — identity, data boundaries, model access, logging, human oversight and supplier assurance.
Read article
AI risk assessment: how to assess an AI system before it ships
A repeatable AI risk assessment covering purpose, data, model behaviour, integration, human oversight, failure modes and evidence.
Read article
ISO 42001 vs ISO 27001: how the two management systems interlock
What each standard covers, where they overlap, and how to run one integrated management system instead of two parallel programmes.
Read article