AI Ethics Cheat Sheet

Responsible AI use begins before you open a tool. Decide whether the task is appropriate, which information you can share and who can review the result. Accuracy, bias and privacy are useful starting points, alongside transparency, intellectual property and human oversight.

Accuracy is the first challenge. AI models can hallucinate, presenting false information with high confidence. Professionals must verify every claim, date, and figure. You are responsible for the truthfulness of the content you share, regardless of whether it was assisted by AI. Accuracy checks against original sources are necessary.

Bias is inherent in the datasets used to train AI. These models can inadvertently replicate societal prejudices regarding race, gender, age, and ability. We must actively look for these biases in AI outputs and correct them. In nonprofit and healthcare settings, this is especially critical to support equitable treatment and representation.

Data handling varies by provider, product, settings and agreement. Inputs may be retained or used for training under some arrangements. Do not assume a tool is private because it has a paid plan. Check your organization’s approval process before sharing any sensitive information.

Never enter PHI or confidential data into non-approved tools. Patient information, donor records, internal personnel files and proprietary plans all need appropriate safeguards. Removing a name may not remove identifying context; use wholly fictional material for learning exercises.

For healthcare work, follow HIPAA requirements and organizational policies. A vendor’s security claim is not the same as approval for your particular task. Ask the appropriate privacy, security or legal team which tools, data and workflows are authorized before proceeding.

For nonprofits, donor data and confidential strategy documents must be handled with care. Privacy settings and data handling depend on the specific provider and their terms of service; a 'privacy toggle' or 'opt-out' should not be interpreted as an approval for handling confidential data. Before using a new tool, consult your organization's security and IT policies. When in doubt, use fictionalized data or high-level summaries that do not reveal sensitive details.

Transparency builds trust. When you use AI to generate content, especially in a professional or educational context, it is often best practice to disclose its use. This does not mean you need a disclaimer on every email, but for significant reports, articles, or creative works, a simple statement can be helpful.

Consider a statement such as: 'Drafting assistance provided by AI, with final review and editing by [Your Name].' This acknowledges the tool while affirming your role as the human authority. In educational settings, being open about AI use helps set expectations for students and staff, fostering a culture of honesty and responsible innovation.

AI should never be the final decision-maker. Whether it is screening job applications, determining eligibility for nonprofit services, or suggesting clinical workflows, a human must always provide the final review and sign-off. This is known as the 'Human-in-the-Loop' approach.

Automated systems lack the ability to understand context, empathy, and complex ethical trade-offs. If an AI suggests a course of action that feels wrong, trust your professional intuition. The AI is a calculator for language and patterns; you are the one with the lived experience and judgment required to make sound decisions.

Intellectual property rights and ownership of AI-assisted work depend on the circumstances and applicable law. Follow organizational policies and ask a qualified legal adviser when rights matter. Do not assume generated text or imagery is yours to use without restriction.

Check whether you have permission to upload source material and reuse the result. Avoid asking a tool to imitate a distinctive protected work. Similarity checks may flag issues but do not establish legal clearance. Keep a record of approved sources and review requirements.

Use this table as a quick reference for daily AI use in your organization. These rules are designed to protect your data and your reputation.

Before starting a task with AI, ask yourself these five questions. If you answer 'No' to any of them, you may need to reconsider your approach or consult with your supervisor.

When the answers are uncertain, pause the task and ask an appropriate colleague. If you discover an error after sharing AI-assisted work, correct it through your organization’s normal process and tell the people who need to know. This guide is educational, not medical, legal or compliance advice.