Glossary
Updated
What is AI ethics?
AI ethics asks whether an AI use is right, not only whether it's legal. It deals with questions such as who could be treated unfairly, whether people know when AI is involved, who is accountable when it goes wrong, and when a person should make the final call. For an organization, AI ethics usually takes the form of a short set of principles, adopted by leadership, that set the standard for every AI decision.
Which frameworks do organizations draw on?
Corporate principles often draw on three international sources. The OECD AI Principles, adopted in May 2019 and updated in May 2024, were the first intergovernmental standard on AI. The EU High-Level Expert Group's Ethics Guidelines for Trustworthy AI, presented in April 2019, set out seven key requirements, including human agency and oversight, transparency, fairness and accountability, and shaped the EU AI Act. UNESCO's Recommendation on the Ethics of Artificial Intelligence was adopted by all 193 member states in November 2021.
How does ethics relate to governance and compliance?
The three overlap but do different jobs. Ethics sets what the organization believes is right. Compliance covers what the law requires. AI governance is the process that applies both to real decisions: who reviews a new AI use, against which criteria, and who signs off. Principles with no process behind them don't change what gets built.
The EU AI Act turned several ethical principles into legal duties, such as human oversight of high-risk systems and telling people when they're talking to a chatbot. Most ethical questions still sit outside the law, which is why organizations write their own principles.
How do you put principles into practice?
Turn each principle into questions reviewers can answer and evidence they can check. Fairness becomes a requirement to test outcomes across groups before launch. Transparency becomes a rule that customers are told when they're dealing with AI. Human oversight becomes a named person who can stop a system. Build those questions into the review every new AI use goes through, so the principles are applied before launch. Many organizations also give an ethics board or Responsible AI team the authority to escalate or block a use. For higher-stakes uses, an AI system impact assessment records who could be affected, and how.
In Enzai, those questions become part of the intake and review each new AI use goes through, so the answers and the decision stay on the record for each use. See Enzai for Responsible AI leaders.
Example
A health insurer's AI principles include no automated denials without human review. When a team proposes a model that flags claims for denial, the intake review asks how a person will check each flag before a claim is denied, and whether outcomes differ by age or ZIP code. The team adds a review step and a quarterly fairness test. The Responsible AI lead approves the use on those conditions, and both are recorded with the decision.