Privacy on English AI Terms Dictionary

Data Minimization

Sat, 18 Jul 2026 10:20:32 +0000

Definition

Data minimization is a core privacy principle requiring organizations to limit data collection to what is adequate, relevant, and limited to what is necessary. In AI, this means designing models that do not require excessive personal information to function accurately. It reduces privacy risks, limits exposure during breaches, and ensures compliance with regulations like GDPR by preventing the accumulation of unnecessary sensitive data.

Summary

The principle of collecting and processing only the personal data that is strictly necessary for a specific purpose.

Right to be Forgotten

Sat, 18 Jul 2026 10:20:32 +0000

Definition

The right to be forgotten enables users to demand the removal of their personal information from databases and AI training sets. Implementing this in machine learning is challenging because models may have memorized patterns from deleted data. Techniques like machine unlearning are being developed to remove the influence of specific data points without retraining the entire model from scratch.

Summary

A legal concept allowing individuals to request the deletion of their personal data held by organizations.

Resisting AI

Sat, 18 Jul 2026 10:14:07 +0000

Definition

Resisting AI refers to methods used by individuals or entities to avoid being influenced, tracked, or classified by AI algorithms. This includes adversarial attacks on perception systems, privacy-preserving data obfuscation, or behavioral changes designed to break predictive models. While often associated with malicious evasion, it also encompasses legitimate privacy advocacy and robustness testing against algorithmic bias or surveillance.

Summary

Strategies and techniques employed to evade detection, manipulation, or control by artificial intelligence systems.

Machine unlearning

Sat, 18 Jul 2026 10:06:25 +0000

Definition

This technique addresses privacy regulations like GDPR’s ‘right to be forgotten’ by allowing models to forget specific user data while retaining general knowledge. It aims to approximate the performance of a model that was never trained on the excluded data. Methods range from exact removal algorithms to approximate gradient-based updates, ensuring compliance and security without the prohibitive computational costs associated with full model retraining.

Summary

Machine unlearning is the process of removing specific data points or their influence from a trained model without retraining it from scratch.

Local Llm

Sat, 18 Jul 2026 10:05:29 +0000

Definition

Running a Local LLM involves deploying open-weight models directly on consumer-grade hardware such as PCs, Macs, or local servers. This approach eliminates reliance on third-party API providers, ensuring complete data privacy since sensitive information never leaves the user’s device. While it requires sufficient computational resources like RAM and GPU memory, advancements in model quantization allow even smaller devices to run capable models. It is ideal for developers and organizations requiring strict compliance, low latency, or operation in disconnected environments.

GDPR Compliance

Sat, 18 Jul 2026 09:58:48 +0000

Definition

GDPR compliance refers to the legal and technical measures AI developers must implement to protect personal data of individuals in the European Union. For AI systems, this involves principles like data minimization, purpose limitation, and the right to explanation for automated decisions. Ensuring compliance requires robust data governance frameworks, transparent algorithms, and mechanisms for users to access, correct, or delete their data, thereby mitigating risks of bias and unauthorized surveillance in AI deployments.

Differential Privacy

Sat, 18 Jul 2026 09:55:28 +0000

Definition

Differential privacy provides strong privacy guarantees by adding calibrated statistical noise to query results or model parameters. It quantifies the maximum amount of information leakage about any single record in a dataset. This technique is crucial for protecting sensitive user information in machine learning pipelines, allowing organizations to derive useful insights from data while maintaining strict confidentiality standards against re-identification attacks.

Summary

A rigorous mathematical framework that ensures the inclusion or exclusion of any single individual’s data does not significantly affect the outcome of an analysis.

Differentially private stochastic gradient descent

Sat, 18 Jul 2026 09:55:28 +0000

Definition

DP-SGD is a variant of Stochastic Gradient Descent designed to protect the privacy of training data. It works by clipping the contribution of each sample’s gradient to limit sensitivity, then adding Gaussian noise scaled to the privacy budget before updating model weights. This process ensures that the final model does not memorize specific training examples, making it resistant to membership inference attacks while maintaining reasonable utility.

Summary

An optimization algorithm that modifies standard SGD by clipping gradients and adding noise to ensure the trained model satisfies differential privacy constraints.

Consent

Sat, 18 Jul 2026 09:51:40 +0000

Definition

In AI ethics, consent refers to the voluntary and informed permission granted by users or subjects before their personal data is collected, stored, or utilized in machine learning models. It requires transparency regarding how data will be used, potential risks, and the right to withdraw permission. Valid consent is a cornerstone of privacy regulations like GDPR, ensuring that individuals maintain agency over their digital footprint and protecting them from non-consensual surveillance or exploitation by AI systems.

Anonymization

Sat, 18 Jul 2026 09:45:50 +0000

Definition

Anonymization involves modifying data so that it can no longer be associated with a specific individual without additional information. This technique is critical in machine learning when handling sensitive personal data, ensuring compliance with regulations like GDPR. Methods include generalization, suppression, and noise addition. While it enhances privacy, effective anonymization must balance utility with risk, as re-identification attacks can sometimes reverse the process if not properly implemented.

Summary

The process of removing personally identifiable information from datasets to protect individual privacy.

Federated Learning

Sat, 18 Jul 2026 09:40:59 +0000

Definition

Federated learning enables organizations to collaboratively train AI models without sharing sensitive raw data. Instead of centralizing information, the model is sent to local devices where it learns from local data, and only model updates (gradients) are transmitted back to a central server for aggregation. This enhances privacy and security, making it ideal for healthcare and finance applications where data sovereignty is paramount.

Summary

Federated learning is a distributed machine learning approach that trains models across decentralized devices while keeping data local.

Data Protection

Sat, 18 Jul 2026 09:40:26 +0000

Definition

Data Protection encompasses legal, technical, and organizational measures designed to secure personal and proprietary data against breaches and misuse. In AI, this includes implementing encryption, access controls, and anonymization techniques to comply with regulations like GDPR. It ensures that training data and user interactions remain private and secure, fostering trust and ethical responsibility in AI systems.

Summary

The process of safeguarding sensitive information from unauthorized access, corruption, or theft throughout its lifecycle.

Local

Sat, 18 Jul 2026 09:33:48 +0000

Definition

In artificial intelligence, ’local’ typically denotes operations performed directly on a user’s hardware, such as a laptop or smartphone, without relying on remote servers. This approach enhances data privacy and reduces latency, as sensitive information does not leave the device. It is increasingly important for edge computing applications where real-time decision-making is critical and network connectivity may be unreliable or non-existent.

Summary

Refers to processing or storing data on a specific device rather than in a centralized cloud environment.