AI ART Discover · Learn · Create
AI Art > Glossary > Differential Privacy
Browse Full AI Glossary
D AI Glossary Term • Term 59 of 224
Plain English Definition

What does “Differential Privacy” mean?

Core Definition

A privacy technique that adds statistical noise so individual records cannot be reliably inferred from outputs.

Operational Mechanism

How Differential Privacy Works

1

Private records

2

Compute a statistic

3

Add calibrated random noise

4

Share the noisy result

5

No single person's record can be pinned down.

Real-World Scenario

Practical Example

A phone maker wants to learn which emoji are popular. Each device adds random noise before sending data, so the company sees accurate totals across many users but cannot tell what any one person used.

Core Significance

Why It Matters

It offers a mathematical privacy guarantee rather than relying on removing names. More privacy means more noise and less accuracy.

Expert Q&A

Common Questions About Differential Privacy

Q: What does the privacy budget mean?

A number, often called epsilon, that limits how much any one person's data can influence the output. Smaller means stronger privacy.

Q: Is anonymization the same?

No. Removing names can still allow re-identification, while differential privacy bounds what can be learned about any individual.

Q: Where is it used?

Government statistics such as the US Census and in some product analytics and model-training settings.