What is local differential privacy and how is it different from global DP?

• A. Local DP adds noise at the data source before sharing; Global DP adds noise during analysis or aggregation on centralized data
• B. Global DP adds noise at the data source before sharing; Local DP adds noise during analysis
• C. Local DP adds noise after aggregation
• D. Local DP and Global DP are the same

Answer: (A)

Local differential privacy perturbs each individual's data at the source before it leaves their device, so the data collector receives only already-noised values and never sees the raw data. Global (centralized) differential privacy assumes a trusted collector that gathers raw data and protects privacy by adding noise to query results or analyses performed on that centralized dataset. The difference lies in where the noise is added and the level of trust required: local DP provides privacy without trusting the data collector, while centralized DP relies on a trusted repository and protects privacy through noisy outputs. The other statements misplace where noise is applied or imply they’re the same.