AI-ordlista
Den kompletta ordlistan över AI
Data Anonymization
Irreversible process of removing or transforming personal identifiers to prevent individual identification, in compliance with GDPR regulations. Effective anonymization must ensure that data can no longer be linked to a specific person through any reasonably available means.
Pseudonymization
Personal data processing technique where direct identifiers are replaced with pseudonyms, allowing controlled re-identification through specific permissions. This method reduces risks while maintaining certain analytical functionalities in AI systems.
Homomorphic Encryption
Advanced cryptographic method allowing computations to be performed directly on encrypted data without prior decryption. This revolutionary technology enables secure processing of sensitive information in untrusted environments.
Privacy by Design
Proactive approach integrating privacy considerations into the design of AI systems from the beginning, rather than as a retroactive addition. This principle requires data protection to be a fundamental component of system architecture and business processes.
Data Minimization
Fundamental data protection principle limiting collection, processing, and retention to information strictly necessary for specified purposes. This practice reduces attack surface and potential exposure of personal data in AI systems.
Federated Learning
Distributed learning paradigm where models train on decentralized local data without data leaving their original environment. This approach preserves privacy while enabling training of performant models on heterogeneous datasets.
Zero-Knowledge Proof
Cryptographic protocol allowing one party to prove the truth of a statement to another party without revealing any information beyond the statement's validity. This technology is particularly useful for identity verification and authentication in distributed systems.
Secure Multi-Party Computation
Cryptographic protocol enabling multiple parties to jointly compute a function on their private inputs without revealing these inputs to each other. This technique facilitates collaboration on sensitive data while preserving each contributor's confidentiality.
Right to be forgotten
Fundamental right allowing individuals to request the deletion of their personal data when it is no longer necessary for the purposes for which it was collected. This right poses complex technical challenges to deep learning-based AI systems.
Privacy Engineering
Applied discipline integrating system engineering principles with privacy requirements to design, develop and deploy privacy-respecting systems. This systematic approach combines tools, methods, and best practices for data protection.
Algorithmic Accountability
Principle requiring organizations to be responsible for decisions made by their AI systems, with mechanisms for transparency, explainability, and recourse. This accountability involves documenting decision-making processes and the ability to justify algorithmic outcomes.
Synthetic Data Generation
Process of artificially creating data statistically similar to real data but containing no personally identifiable information. This technique enables the development and testing of AI models while bypassing privacy restrictions.
Consent Management Platform
Technology system centralizing the collection, management, and documentation of user consent for processing their personal data. These platforms ensure regulatory compliance and facilitate the exercise of data subjects' rights.
Data Governance Framework
Organizational structure defining policies, standards, and procedures to manage data as a strategic asset while ensuring its protection. This framework establishes responsibilities, validation processes, and mechanisms for quality and security control.
Privacy-Preserving Machine Learning
Set of techniques and algorithms enabling the training and deployment of machine learning models without compromising the confidentiality of training data. These methods include federated learning, encryption, and noise-based approaches.
Data Protection Impact Assessment
Systematic process for assessing risks to individuals' rights and freedoms related to the processing of their personal data in AI systems. This assessment is mandatory for high-risk processing under the GDPR.