A Plain-English Glossary for AI, Privacy, and Legal Tech

The rules, practices, oversight structures, and accountability measures used to decide how AI systems should be developed, deployed, monitored, and challenged.

A technical system that uses data, rules, models, or statistical methods to support or make decisions. In legal and public-institution settings, the key question is often whether a person can tell that the system was used and contest a harmful result.

A pattern where a technical system produces unfair or unequal outcomes. Bias can come from training data, design choices, deployment context, institutional practices, or the way results are interpreted by people.

The set of rights, duties, and design choices that govern how information about people is collected, used, shared, retained, secured, and deleted.

Software, data systems, and digital tools used in legal work, court administration, compliance, legal aid, dispute resolution, document preparation, research, or client services.

The practical ability to understand, protect, and exercise legal rights. This includes access to information, language support, counsel, fair procedures, usable forms, understandable notices, and meaningful ways to challenge errors.

Technology work organized around public needs, rights, equity, accountability, and institutional responsibility rather than only private efficiency or profit.

The process of testing how a model performs, where it fails, and whether it is appropriate for a particular use. Evaluation should consider not only accuracy but also error patterns, affected groups, context, and consequences.

A tool or process that estimates the likelihood of a future event or harm. In legal systems, risk assessments require special caution because scores can influence liberty, supervision, benefits, housing, employment, or access to services.

A structured review of how a project or system collects and uses personal data, what risks it creates, and what safeguards or alternatives should be in place.