privacy

Protecting individuals from unauthorized data collection, inference, and use in AI systems. Includes data minimization, anonymization, and preventing model memorization.

Syntax

ai-ethics

privacy_by_design: collect minimum data, anonymize, prevent leakage

Example

ai-ethics

// Privacy best practices for AI:

// 1. Data minimization — collect only what is needed:
const userData = { id: hash(email), age_group: "25-34" }; // not exact age

// 2. Differential privacy — add noise to protect individuals:
// Use libraries like Google DP Library, Apple DP

// 3. Prevent training data leakage:
// Test: does model memorize PII from training data?
// Use: membership inference attack detection

// 4. GDPR/CCPA compliance:
// Right to deletion — can you retrain without a user's data?