Using AI Ethically at Work: A Simple Personal Policy

Most people using AI at work are making it up as they go. There is no company policy, no guidance from their manager, and no clear line between smart use and risky behaviour. This article gives you a one-page personal policy for using ai ethically at work, covering the three areas where problems actually arise: disclosure, accuracy checking, and confidentiality.

Why you need a personal AI ethics policy

Your employer probably does not have an AI usage policy yet. A few large Kenyan organisations (banks, telcos, some NGOs) have started drafting them, but most businesses have not. This means the ethical decisions about how to use AI at work currently fall on you.

This is not a theoretical concern. Real problems are already happening in Kenyan workplaces.

A marketing associate submits an AI-generated report as their own work. Their manager discovers it, and trust is broken, not because AI was used, but because it was not disclosed.

An accountant asks ChatGPT to summarise KRA tax regulations and includes the output in a client advisory letter. One of the "facts" is wrong. The client files based on incorrect information.

A human resources officer pastes employee performance review details into an AI tool to help write summaries. Those personal details are now stored on external servers.

In each case, the person was trying to be more productive. The problem was not AI use itself. It was the absence of clear personal rules about disclosure, checking, and data handling. A simple policy prevents these scenarios.

The one-page policy

Print this out. Put it where you work. Adjust the specifics to your role, but keep the three sections.

Section 1: Disclosure

Rule: I will be transparent about when and how I use AI in my work.

• If I use AI to draft content that carries my name or my company's name (reports, proposals, emails to clients, social media posts), I will either disclose the AI's role or review and edit the output thoroughly enough that I can stand behind every word as my own.
• If my manager or client asks whether AI was involved in producing my work, I will answer honestly.
• I will not present AI-generated analysis, research, or recommendations as if I personally conducted the analysis or research.
• If I am unsure whether disclosure is expected in a specific situation, I will disclose. Transparency rarely creates problems. Concealment often does.

What this means in practice: If you use ChatGPT to draft a client proposal, you have two ethical options. Either tell the client ("I used AI to help draft this, and I have reviewed and edited it for accuracy") or edit it so thoroughly that you have genuinely authored the final version. What is not acceptable is submitting raw AI output as your original work, especially when the recipient is making decisions based on it.

Section 2: Accuracy checking

Rule: I will verify every factual claim, number, and recommendation that AI generates before passing it along.

• I will not include statistics, legal references, or technical specifications from AI output in my work without verifying them against a primary source.
• I will treat AI outputs as first drafts, not final answers. Every claim needs a human check before it leaves my desk.
• If I cannot verify a specific AI-generated claim and I need to include it, I will flag it as unverified.
• I will be especially cautious with AI outputs about Kenyan regulations, tax rules, legal requirements, and local market data, as AI models are often less accurate about Kenya-specific information than about widely documented topics.

What this means in practice: AI tools confidently present wrong information. They invent case law, fabricate statistics, and cite regulations that do not exist. In a Kenyan work context, this is particularly risky for anything involving KRA requirements, county regulations, or sector-specific compliance. If ChatGPT tells you that a certain tax exemption applies to your client's situation, verify it on the KRA portal or with a qualified tax professional before acting on it.

Section 3: Confidentiality

Rule: I will not share confidential, personal, or sensitive information through AI tools without appropriate safeguards.

• I will not paste client personal data (names, ID numbers, phone numbers, financial details) into AI tools on free-tier plans.
• I will not input proprietary company information (pricing strategies, unreleased plans, internal financial data) into AI tools unless using an enterprise plan with verified data protection terms.
• I will not input employee personal information (salaries, performance reviews, medical records, disciplinary matters) into AI tools.
• I will not use personal AI accounts for work involving sensitive data. If work requires AI, I will use work-approved tools and accounts.
• I will review my AI tool privacy settings quarterly to ensure they have not changed.

What this means in practice: If you need AI help writing a salary benchmarking analysis, you can ask general questions ("What is the typical salary range for a marketing manager in Nairobi?"). You should not paste your company's actual salary spreadsheet into the tool. The general question is safe. The specific data is not.

Applying the policy to common work scenarios

Scenario 1: Drafting a report for your manager

You use AI to help outline and draft a monthly sales report. Disclosure: If your workplace culture treats reports as your personal work product, edit the draft until it truly reflects your analysis. If your manager is aware that team members use AI tools, a simple mention is sufficient. Accuracy: Verify every sales figure against your actual data. AI does not have access to your company's real numbers and may generate plausible but incorrect figures. Confidentiality: Do not paste actual customer names or transaction details into the AI tool. Use anonymised or aggregated data.

Scenario 2: Writing a client email

A client has asked a complex question, and you want AI help crafting a clear response. Disclosure: The client expects a response from you, and what matters is that the information is correct. Edit the response to ensure accuracy and your professional voice. Accuracy: If the email contains specific claims (timelines, costs, regulations), verify each one. Confidentiality: Do not paste the client's original email (which may contain their sensitive information) into an AI tool.

Scenario 3: Preparing interview questions for a job candidate

You ask AI to help generate interview questions for a specific role. Disclosure: There is no expectation to disclose that interview questions were AI-assisted, as long as the questions are appropriate and relevant. Accuracy: Review each question for relevance to the role and compliance with Kenyan employment law. Ensure no questions violate anti-discrimination provisions. Confidentiality: Do not paste the candidate's CV or personal details into the AI tool when generating questions.

Scenario 4: Analysing data in a spreadsheet

You want to use AI to identify patterns in a sales dataset. Disclosure: If the analysis informs business decisions, be transparent about the methodology. Accuracy: Cross-check AI-identified patterns against the actual data. AI may see patterns that do not exist or miss obvious ones. Confidentiality: If the dataset contains customer information, anonymise it before using AI, or use a tool with enterprise data protection.

What to do when your employer's policy conflicts with yours

Some employers may explicitly encourage AI use without clear guidelines on disclosure or data protection. Others may ban AI tools entirely, which may not be practical. If your employer has a policy, follow it, but supplement it with your personal rules where the company policy is silent.

If there is a genuine conflict (your employer asks you to do something with AI that you believe is ethically wrong, like using AI to generate fake reviews or fabricate credentials), that is a professional ethics issue that goes beyond AI. Trust your judgment and, if necessary, seek guidance from a professional body or legal counsel.

Keeping the policy current

AI tools change rapidly. Privacy settings get updated, new features launch, and workplace norms shift. We review our own AI ethics practices every quarter. You do not need a formal review process. Just re-read your one-page policy every few months and ask: does this still cover the situations I face?

Our AI and Automation for Beginners course walks through responsible AI use with practical exercises, including building and refining your own workplace AI policy.

FAQ

Do I need to disclose AI use if I only used it for brainstorming?

If you used AI to generate ideas and then developed those ideas entirely through your own thinking and writing, disclosure is less critical. The key question is: could someone be misled about the origin of the final work? If the final output is genuinely yours and the AI only sparked an initial direction, most professional contexts would not require disclosure. When in doubt, mention it.

What if everyone on my team uses AI and nobody discloses it?

Common practice does not equal ethical practice. If your team culture normalises undisclosed AI use, you can still choose to be transparent. Over time, teams that are open about AI use tend to develop better practices, like sharing effective prompts and catching AI errors collaboratively. You can model this by casually mentioning your AI use rather than making it a formal announcement.

Can I be fired for using AI at work in Kenya?

Kenyan labour law does not specifically address AI tool use. Whether AI use at work constitutes misconduct depends on your employment contract, any company policies, and the nature of the use. Using AI to be more productive is different from using it to fabricate work products or breach confidentiality. If you follow the disclosure, accuracy, and confidentiality principles in this article, you are on solid ground. If your employer has no AI policy, suggesting one proactively shows initiative rather than risk.

How do I handle AI ethics when freelancing for international clients?

International clients may have their own AI policies. Ask about them during the onboarding process. If they do not have one, the personal policy in this article still applies. Your professional reputation depends on delivering work that is accurate and transparently produced, regardless of where the client is based.