What does “Differential Privacy” mean?
A privacy technique that adds statistical noise so individual records cannot be reliably inferred from outputs.
How Differential Privacy Works
Private records
Compute a statistic
Add calibrated random noise
Share the noisy result
No single person's record can be pinned down.
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.
Why It Matters
It offers a mathematical privacy guarantee rather than relying on removing names. More privacy means more noise and less accuracy.
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.