2 docs tagged with "privacy budget"

DP fine-tuning: differential privacy bounds any single sample's influence on me — but ε isn't zero, and it isn't zero leakage

In one sentence clip per-sample gradients + add noise) gives a provable property ε is not zero — it's "bounded leakage," not "zero leakage"; and more private usually means lower utility. Don't read "DP was applied" as "private" — look at what ε is, and whether it protects a sample or a user.

DP in-context learning: put differential privacy on the private examples in your prompt, not on training

In one sentence either it splits the private examples into disjoint subsets and does a noisy aggregation over my outputs across those subsets before answering, or it uses the private data to generate a batch of synthetic examples with an (ε, δ) guarantee that replace the real ones. What it can bound is any single private example's influence on the answer — not me introspecting the prompt. Two boundaries up front: it protects the examples, not the query itself; and ε isn't zero, so out-of-aggregation side channels still leak.