Differential Privacy
Differential privacy is a mathematical technique that adds carefully calibrated noise to data or results so that nothing can be reliably learned about any single individual.
What is Differential Privacy?
It gives a measurable privacy guarantee and is used by statistics agencies and technology firms, and in training AI models on sensitive data. Stronger privacy reduces accuracy.
How it works
Random noise scaled to a privacy budget, epsilon, is added to queries, data or training updates.
← All Artificial Intelligence & Superintelligence definitions · ← Back to the A–Z definitions