Definition
Data ethics covers the norms and decisions around data that go beyond pure technical possibility. It includes purpose limitation, data minimization, quality, provenance, access control, and how data impacts people when used in AI systems.
Why it matters
- Trust: clients and users expect careful handling of sensitive information.
- Quality: bad data produces bad models and misleading outputs.
- Risk: unethical data practices create legal, security, and reputational problems.
How it works
Purpose -> minimize -> secure -> document -> monitor -> correctPractical controls include data governance, retention rules, access management, and documentation of sources and consent where relevant.
Practical example
A professional services tool keeps client data separated by matter, logs access, and avoids training general models on sensitive client documents without explicit governance approval.
Common questions
Q: Is data ethics the same as privacy law?
A: No. Privacy law is a baseline. Data ethics also covers legitimacy, expectations, and potential harm.
Q: Can ethical data use still be high-performance?
A: Yes. Strong governance and clean data often improve reliability and reduce downstream risk.
Related terms
- AI Governance Framework — policies and controls
- AI Risk Management — data risks as part of the risk register
- Bias Mitigation — fairness depends on data choices
- AI Documentation Requirements — document data provenance and limits
References
OECD (2019), OECD Principles on Artificial Intelligence.