Differential privacy

A differentially private mechanism releases a statistic with enough noise that any single individual’s presence changes the output distribution by at most a factor of e^epsilon. Choosing epsilon is a governance decision: smaller epsilon = stronger privacy = noisier results; the sensitivity of the query (how much one record can move the result) scales the noise.

Confium ships Laplace and Gaussian mechanisms in confium-privacy (confium privacy dp --value ... --sensitivity ... --epsilon ...). Track budgets across releases — privacy loss composes.