Glossary of terms in EU AI Act
This glossary is not exhaustive, and there may be other terms that are used in the EU AI Act that are not defined here. However, it provides a good starting point for understanding the key terminology used in the regulation.
Artificial Intelligence (AI) System
A computer-based system designed to perform tasks that typically require human intelligence, such as visual perception, speech recognition, decision-making, and language translation.
High-Risk AI System
AI applications that pose significant risks to health, safety, or fundamental rights of individuals, often subject to stricter regulatory oversight.
Unacceptable AI System
AI applications considered to create unacceptable risks and are typically prohibited by law, such as certain types of social scoring systems.
General Purpose AI System
An AI system with broad capabilities that can be applied to a wide range of tasks across different domains.
AI Regulatory Sandbox
A controlled environment where companies can test innovative AI technologies under regulatory supervision, often with temporary exemptions from certain rules.
Notified Body
An organization designated by a country to assess the conformity of certain products before they can be placed on the market.
National Competent Authority
Government bodies responsible for implementing and enforcing AI-related regulations at the national level.
AI Office
A dedicated governmental or organizational unit responsible for overseeing AI-related policies, implementation, and governance.
Post-Market Monitoring System
A process for continuously assessing and addressing risks associated with AI systems after they have been deployed.
AI Literacy
The ability to understand, use, and critically evaluate AI technologies and their impacts on society.
Conformity Assessment
The process of evaluating whether an AI system meets specified requirements, standards, or regulations.
Placing on the Market
The act of making an AI system or model available for sale or use for the first time in a particular market or region.
More glossary terms related to AI in general
Algorithmic Bias
Systematic errors in AI systems that can result in unfair outcomes for certain groups, often based on characteristics like race, gender, or age.
Automated Decision-Making
The process by which decisions are made by AI systems without human intervention.
Biometric Data
Personal data related to an individual's physical, physiological, or behavioral characteristics.
Data Minimization
The principle of limiting data collection and processing to only what is necessary for a specific purpose.
Explainable AI
AI systems designed to provide understandable explanations for their decisions and actions.
Federated Learning
A machine learning technique that trains algorithms across multiple decentralized devices or servers without exchanging raw data.
Human-in-the-Loop
A model that requires human interaction in the decision-making process.
Model Drift
The degradation of an AI model's performance over time due to changes in data or environment.
Privacy-Preserving AI
Techniques used to develop AI systems that protect individual privacy while maintaining functionality.
Responsible AI
The practice of designing, developing, and deploying AI systems in an ethical, transparent, and accountable manner.
Risk-Based Approach
A method for categorizing and regulating AI systems based on their potential risk level.
Synthetic Data
Artificially generated data that mimics real-world data, often used for training AI models.
Transparency by Design
The principle of incorporating transparency into AI systems from the earliest stages of development.
AI Ethics Committee
A group of experts providing guidance on ethical issues related to AI development and deployment.
AI Impact Assessment
A systematic evaluation of the potential effects of an AI system on individuals and society.
Algorithmic Accountability
The obligation to explain and justify decisions made by AI systems.
Data Governance
The overall management of data availability, usability, integrity, and security within an organization.
Model Interpretability
The degree to which an AI model's decision-making process can be understood by humans.
Robustness Testing
The process of evaluating an AI system's performance under various conditions, including adversarial attacks.
Trustworthy AI
AI systems that are lawful, ethical, and technically robust.
