GLOSSARY > DIFFERENTIAL PRIVACY
Differential Privacy
A mathematical framework for guaranteeing that the output of an algorithm cannot be used to reverse-engineer individual data points in the training set.
Differential Privacy (DP) introduces controlled cryptographic noise during the training or querying of AI models. This statistical noise ensures that the inclusion or exclusion of any single record does not significantly alter the model's output. For Private AI deployments handling PII or PHI, DP is a non-negotiable safeguard against model inversion and membership inference attacks.
Explore More