Your Workplace Uses AI Where Is Your AI Policy

AI is already inside the workplace, even when no one has approved it.
An employee uses a chatbot to draft a client email. A supervisor asks an AI tool to summarize complaints. A manager uploads a spreadsheet to find patterns. Someone pastes a contract clause into a free tool because it is faster than waiting for review.
None of this means the organization is careless. It means the tools moved faster than the rules.
That is why small and mid-sized organizations need a practical workplace AI policy now. Not a 50-page manual. Not a technical document that only one person understands. A clear one-page policy that tells employees what they can do, what they cannot do, when they need approval, and who is accountable when something goes wrong.
This closing article in the October Series: AI at Work Without Losing the Human brings the issue back to leadership. AI adoption is not just a technology choice. It is a trust decision.

Why a one-page AI policy is no longer optional
Generative AI is changing everyday work. The International Labour Organization has reported that about one in four workers globally are in occupations with some exposure to generative AI. The ILO also stresses that job transformation is more likely than full job replacement in many roles.
That distinction matters.
For many organizations, the near-term risk is not that AI will replace everyone. The immediate risk is that people will use AI without clear guidance. They may expose confidential data, rely on inaccurate output, create copyright problems, or use tools in ways that undermine fairness.
Jamaica’s Future of Work agenda makes this even more relevant. The national conversation now includes digital-economy policy, remote-work standards, AI-enabled industries, productivity, accountability, worker wellbeing, cybersecurity, and digital responsibility. Those priorities are not abstract. They show up in daily decisions made by managers, employees, contractors, and service providers.
A simple AI policy helps answer basic questions:
Can staff use AI to write first drafts?
Can customer information be entered into an AI tool?
Who checks AI-generated content before it reaches the public?
Can AI support hiring or performance decisions?
What happens if someone discovers an AI-related error?
Without answers, convenience becomes the policy by default.
That is why the phrase AI POLICY: BEFORE CONVENIENCE BECOMES RISK. belongs at the center of this conversation.
What responsible AI governance should cover
NIST’s work on responsible AI risk management gives organizations a useful starting point. The core idea is simple: identify the risks, measure or assess them where possible, manage them with clear controls, and keep monitoring over time.
For a small or mid-sized organization, that does not require a large compliance department. It requires plain rules.
A good one-page workplace AI policy should cover these areas.
Approved uses
Start by naming the safe, useful ways employees may use AI.
This removes guesswork and encourages better adoption. Most organizations can allow AI for low-risk support tasks, such as:
Drafting internal notes or outlines
Summarizing non-confidential material
Creating first drafts of routine communications
Brainstorming training ideas
Rewriting text for clarity
Producing checklists or meeting preparation notes
The key is to make clear that AI output is a draft, not a final authority.
When employees know the approved uses, they are less likely to hide their use of AI. They are also more likely to ask for help when a use case feels uncertain.
Prohibited uses
Every policy needs clear red lines.
Some AI uses create too much risk for ordinary employee discretion. These should be banned unless and until the organization creates stronger controls.
Common prohibited uses include:
Entering passwords, private credentials, or banking details into public AI tools
Uploading confidential contracts, employee files, customer records, or medical information into unapproved systems
Using AI to impersonate a person
Creating misleading content, fake evidence, or false reviews
Using AI to bypass cybersecurity, privacy, or HR procedures
Allowing AI to make final decisions about hiring, discipline, promotion, pay, or termination
Prohibited uses should be short, specific, and easy to remember. If the list is vague, staff will interpret it differently.
Confidential information
Confidential information needs special protection because generative AI tools often rely on user inputs to process a request. Depending on the tool and its settings, data may be stored, reviewed, or used in ways the organization does not expect.
A one-page policy should tell employees not to enter sensitive information into public or unapproved AI tools.
That includes:
Customer names, contact details, account details, and complaints
Employee files, payroll information, disciplinary records, and health-related information
Financial data, pricing strategies, contracts, and supplier terms
Internal passwords, access codes, system details, and cybersecurity information
Unreleased plans, reports, proposals, and board materials
If an organization approves a secure AI tool for sensitive work, the policy should name who is allowed to use it and for what purpose.

Use the four-category framework
A one-page policy becomes easier to follow when every AI use falls into one of four categories: Allowed, Restricted, Prohibited, or Human Approval Required.
Category | What it means | Examples |
Allowed | Employees may use AI without special approval if they follow the policy. | Drafting internal outlines, summarizing public information, improving grammar, creating non-sensitive checklists. |
Restricted | Employees may use AI only with approved tools, approved data, or defined safeguards. | Analyzing internal reports, preparing customer service summaries, translating business content, supporting research. |
Prohibited | Employees must not use AI for these purposes. | Uploading confidential data to public tools, impersonation, final HR decisions, misleading content, cybersecurity misuse. |
Human Approval Required | AI may support the work, but a named person must review and approve before action. | Public statements, legal or regulatory material, hiring support, performance documentation, customer-impacting decisions. |
This framework works because it gives employees a quick sorting tool.
If the task is low risk, it may be allowed. If the task uses internal data, it may be restricted. If the task affects rights, employment, privacy, money, safety, or reputation, it likely needs human approval. If the task involves deception, sensitive data, or bypassing controls, it should be prohibited.
The goal is not to slow useful work. The goal is to prevent avoidable harm.
Build the one-page workplace AI policy
A strong one-page AI policy should answer ten practical questions. Each one can fit into a short section.
1. What tools may employees use?
Name approved AI tools or explain how tools are approved.
If the organization has not approved any tools yet, say so clearly. Allowing “any AI tool” is risky because tools differ in privacy settings, security standards, data use terms, and reliability.
The policy should also say that employees must not connect AI tools to company systems, email accounts, cloud folders, or databases without permission.
2. What uses are approved?
List everyday uses that support productivity without creating major risk.
Keep the wording plain. For example:
Employees may use approved AI tools to draft, summarize, organize, rewrite, and generate ideas for low-risk work, provided no confidential, personal, or sensitive information is entered.
That one sentence gives useful permission and a clear boundary.
3. What information must never be entered?
This section should be direct.
Employees should not enter confidential business information, customer data, employee data, passwords, payment details, legal documents, private correspondence, or sensitive operational information into public or unapproved AI tools.
If the organization handles regulated data, this section should be reviewed by legal, compliance, or a qualified adviser.
4. How should employee and customer data be handled?
AI policies must protect people, not just files.
Customer and employee data should be used only for a lawful, necessary, and approved purpose. Staff should collect and share the minimum amount of data needed. Where possible, they should remove names and identifying details before using approved tools.
The policy should also tell employees to report any accidental data entry into an AI tool.
Silence makes incidents worse. Early reporting helps the organization contain harm.
5. Who checks the facts?
AI can sound confident when it is wrong.
The policy should require human fact-checking before AI-assisted content is used in reports, client communications, public posts, financial material, legal material, HR documents, or operational decisions.
A simple rule works well:
No AI-generated output may be treated as true until a human checks it against reliable sources.
This matters for customer trust, worker wellbeing, and accountability.
6. How will copyright and ownership be handled?
AI tools can produce text, images, code, or other material that may create copyright or licensing questions.
The policy should require employees to avoid asking AI tools to copy a living artist’s style, reproduce protected text, imitate a competitor’s materials, or generate content based on copyrighted documents the organization does not have permission to use.
Staff should also disclose when AI helped create public-facing content, if the organization’s procedure requires it.
7. When is human review mandatory?
Human review should be required when AI output may affect people, rights, services, safety, employment, money, or reputation.
This includes:
Hiring shortlists or interview questions
Performance notes or disciplinary material
Customer complaints or service decisions
Public communications
Legal, financial, or compliance-related documents
High-impact operational choices
AI can support these tasks, but a person must remain responsible for the final decision.

Hiring and HR decisions need extra care
AI can create serious risk in hiring and HR because these decisions affect livelihoods. Bias, inaccurate assumptions, poor data, and weak oversight can harm applicants and employees.
A one-page AI policy should state that AI may not make final HR decisions.
If AI is used to support HR work, the organization should require:
A human decision-maker
Clear documentation of how the AI tool was used
Review for bias or unfair impact
Protection of applicant and employee data
A way for people to ask questions or challenge an outcome
This aligns with the ILO’s focus on job transformation, worker wellbeing, and human-centered work. It also supports NIST’s focus on accountability, transparency, and harm reduction.
The safest rule is simple: AI can assist HR work, but it cannot replace human judgment.
Accountability cannot be vague
A policy is only useful if people know who owns it.
The one-page version should name the roles responsible for AI governance. In a small organization, that may be the managing director, operations lead, HR lead, IT provider, or data protection officer. In a mid-sized organization, it may be a small AI review group.
The policy should explain:
Who approves AI tools
Who approves restricted uses
Who reviews incidents
Who updates the policy
Who trains employees on the rules
Accountability also means employees should not be punished for reporting honest mistakes early. If someone accidentally enters the wrong information into a tool, the organization needs to know quickly.
A blame-first culture drives AI use underground. A clear reporting process builds trust.
Incident reporting should be simple
AI incidents will happen. The policy should define what counts as an incident and how to report it.
Examples include:
Confidential data entered into an unapproved AI tool
AI-generated content sent externally with false information
A customer or employee harmed by AI-assisted output
A suspected bias issue in hiring, service, or HR use
AI used to impersonate someone
An unapproved AI tool connected to company systems
The reporting route should be short. Employees should know whom to contact, what to send, and how quickly to report.
For example:
Report AI incidents to the assigned manager or data lead within one business day. Include the tool used, the information entered, the output received, who saw it, and any action already taken.
That level of detail helps the organization respond without panic.
The One-Page Workplace AI Policy Starter
Use this starter as a practical first draft. Adapt it to fit the organization’s size, sector, legal duties, and risk level.
Purpose
This policy explains how employees, contractors, and managers may use AI tools at work. It protects confidential information, customer and employee data, fairness, accuracy, copyright, cybersecurity, and organizational trust.
Approved tools
Employees may use only AI tools approved by the organization. New tools must be reviewed before use.
Allowed uses
Employees may use approved AI tools for low-risk tasks such as drafting, summarizing, organizing, rewriting, translating non-sensitive content, and generating ideas.
Restricted uses
AI may be used for internal analysis, customer-related summaries, operational reports, or other higher-risk work only with approved tools, approved data, and manager guidance.
Prohibited uses
Employees must not enter confidential, personal, sensitive, financial, legal, HR, cybersecurity, or customer-identifying information into public or unapproved AI tools. Employees must not use AI to deceive, impersonate, make final HR decisions, bypass controls, or create misleading content.
Human approval required
A human reviewer must approve AI-assisted work before it is used for public communication, customer-impacting decisions, HR decisions, legal or financial material, compliance matters, or any decision affecting a person’s rights, employment, money, safety, or access to services.
Fact-checking
Employees must verify AI output before using it. AI-generated content must not be treated as accurate without review.
Copyright and content use
Employees must not use AI to copy protected material, imitate copyrighted work, or create content from materials the organization has no right to use.
Accountability
Managers are responsible for approved use within their teams. Employees are responsible for following this policy and asking for guidance when unsure.
Incident reporting
Employees must report AI mistakes, data exposure, harmful output, or suspected misuse within one business day to the assigned manager or data lead.
Review
This policy should be reviewed regularly as tools, laws, risks, and business needs change.

Good AI governance is leadership in plain language
AI policy should not begin with fear. It should begin with responsibility.
People will use tools that help them work faster. Leaders have to make sure speed does not weaken privacy, fairness, accuracy, accountability, or trust. That is the real governance challenge for small and mid-sized organizations.
A one-page AI policy will not solve every issue. It will not replace training, legal review, cybersecurity controls, or good management. But it gives everyone a shared starting point.
It tells employees, “You can use AI, but use it responsibly.”
It tells customers, “We are not treating your information casually.”
It tells workers, “Human judgment still matters here.”
It tells leaders, “Technology adoption is now part of organizational trust.”
That is the right place to end this October series: not with a warning against AI, but with a call for better leadership. The organizations that earn trust will not be the ones that avoid AI completely. They will be the ones that adopt it with clear rules, human oversight, and the courage to be accountable.





Comments