16 November 2026
Many organisations are still talking about AI as something they are planning to adopt.
Their employees may tell a different story.
Generative AI tools can be accessed in seconds. AI functionality is increasingly built into familiar workplace software. Suppliers are adding AI capabilities to existing services.
Employees may be using AI to draft emails, summarise documents, research topics, analyse information, create presentations, write code or support customer interactions.
That means the first question for many organisations is no longer:
Should we adopt AI?
It is:
How are we already using it?
For SMEs preparing for 2027, understanding that difference is an important starting point for effective AI Governance.
Start with visibility
You cannot govern something you do not know is happening.
Before creating a complicated AI framework, organisations should understand where AI is already being used.
Which tools are employees accessing?
Which existing applications now contain AI functionality?
What business activities are those tools supporting?
Are free public tools being used or organisation-approved platforms?
Are suppliers using AI within services provided to you?
What outputs are people relying on?
This does not need to begin as a huge discovery exercise.
The objective is to establish enough visibility to understand where the most important uses and risks may sit.
Understand what information is being shared
One of the most immediate AI Governance questions concerns data.
Employees using generative AI tools may enter:
Whether that creates a problem depends on the tool, its terms, the information involved and how it is being used.
But employees cannot make appropriate decisions if nobody has explained what is acceptable.
Organisations therefore need clear guidance around the information that can and cannot be shared with AI tools.
This is not simply an IT issue.
It involves information security, privacy, legal, commercial and reputational risk.
Think about the output as well as the input
AI Governance is not only about what information goes into a system.
It is also about what comes out.
AI-generated information can be inaccurate, incomplete or misleading.
Content can appear convincing even when the underlying answer is wrong.
That becomes more important when AI outputs influence decisions.
Is somebody checking the result?
Does the employee understand the limitations of the tool?
Could the output affect a customer?
Is AI being used to support a low-risk administrative task or a more significant decision?
The appropriate level of oversight should reflect the context and potential impact.
Not every use of AI needs the same controls.
Give employees something practical to work with
Telling employees to “use AI responsibly” is unlikely to provide enough guidance.
People need to understand what responsible use means within their organisation.
Which tools are approved?
Which information can be used?
Which activities require human review?
When should an employee ask for approval?
How should AI-generated information be checked?
What uses are prohibited?
Who should employees speak to if they are unsure?
A practical AI policy or acceptable-use framework can help create consistency.
But a policy is only useful if employees know it exists and understand how to apply it.
AI literacy and awareness therefore form an important part of governance.
Consider the suppliers too
AI can enter an organisation without an employee deliberately choosing an AI tool.
Existing technology suppliers may introduce AI functionality into their platforms.
Service providers may use AI within their own delivery processes.
Software updates may activate new capabilities.
Organisations should therefore consider AI within supplier and technology assessment.
What AI functionality is being provided?
What data does it use?
Where is that data processed?
What decisions or outputs does it influence?
What controls does the supplier provide?
Can the functionality be configured or disabled?
What assurance can the supplier provide?
Supplier governance becomes increasingly important as AI becomes embedded across the technology ecosystem.
Make responsibility clear
Who owns AI Governance?
For some SMEs, the answer may not be one person.
Technology teams may understand the tools.
Information security may understand data risks.
HR may need to consider employee use.
Legal or compliance teams may consider contractual and regulatory implications.
Business leaders understand where AI could create value.
The important thing is that responsibility does not disappear between those functions.
Someone needs oversight of how AI is being adopted and how significant risks are being managed.
Without clear responsibility, AI adoption can develop faster than the organisation’s ability to understand it.
Governance should be proportionate to risk
A common concern about governance is that it will slow innovation.
It can, if it is designed badly.
Requiring the same approval process for every possible use of AI would quickly become impractical.
Using AI to help structure an internal meeting agenda does not necessarily create the same risk as using AI to influence an important customer or employment decision.
Good governance should recognise that difference.
Lower-risk use cases may require straightforward guidance.
Higher-risk applications may require more detailed assessment, oversight, testing or approval.
The objective is not maximum control.
It is appropriate control.
Use recognised frameworks where they help
As organisations develop more formal approaches to AI Governance, recognised frameworks can provide useful structure.
ISO/IEC 42001 provides a management-system framework for the responsible development and use of AI.
It can help organisations consider areas such as responsibilities, risk, objectives, controls, monitoring and continual improvement.
But the same principle applies here as elsewhere in governance.
The standard should support the outcome.
The objective is not simply to achieve certification.
It is to establish an approach that allows the organisation to use AI responsibly and demonstrate that appropriate governance exists around it.
Keep reviewing as AI changes
AI Governance cannot be a one-off project.
Tools will change.
New functionality will appear.
Employees will identify new use cases.
Suppliers will introduce new capabilities.
Risks and expectations will evolve.
Governance therefore needs a process for review.
Are approved tools still appropriate?
Have new uses emerged?
Have incidents or concerns been reported?
Do policies need updating?
Do employees need further guidance?
Have supplier arrangements changed?
Are controls still proportionate to the risks?
The aim is to keep governance aligned with actual AI use rather than allowing the framework and reality to drift apart.
Take Control of AI Without Stopping Innovation
AI presents significant opportunities for organisations.
The answer to risk is not necessarily to avoid those opportunities.
It is to understand them.
Know where AI is being used.
Understand the information involved.
Assess the risks.
Give employees clear guidance.
Review suppliers.
Establish ownership.
Apply controls proportionate to the use case.
Keep reviewing as things change.
Good AI Governance should not answer every new idea with “no”.
It should help the organisation understand when it can confidently say “yes”.
If you are reviewing your approach to AI Governance and responsible adoption, talk to KA2.