AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Provably Protecting Fine-Tuned LLMs from Training Data Extraction while Preserving Utility

Tom Segal, Yuval Elovici, Asaf Shabtai

ICML 2026 · Large language models › Instruction and fine-tuning · approach: Safety alignment after fine-tuning · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
aggregate over relative probability distributions instead of raw probabilities (CP-Delta to CP-Delta_r)CP-Delta · Canary (MathAbstracts)mean exposure: 8.8 -> 3.6Improved
add base-model smoothing (pre-fine-tuning base) to CP-Delta_r (ablation)CP-Delta_r · Canary (MathAbstracts)mean exposure: 3.6 -> 1.4Improved
add base-model smoothing (pre-fine-tuning base) to CP-Delta_r (ablation)CP-Delta_r · PII extraction (Names2IDs)full ID extraction rate (FER): .06 -> <.001Improved
add base-model smoothing (constant-logit base) to CP-Delta_r (ablation)CP-Delta_r · PII extraction (Names2IDs)full ID extraction rate (FER): .06 -> .05Improved
add base-model smoothing (pre-fine-tuning base) to CP-Delta_r (ablation)CP-Delta_r · TTE (MathAbstracts)attack AUC: .80 -> .76Improved
add base-model smoothing to CP-Delta_r (ablation)CP-Delta_r · MathAbstracts99th percentile k_x: .98 -> .004Improved
use relative probabilities instead of raw probabilities (CP-Delta to CP-Delta_r)CP-Delta · MathAbstracts99th percentile k_x: 11.6 -> .98Improved
replace RPD aggregation by raw-probability CP-Delta on the smoothed distributions (ablation)SCP-Delta_r · Canary (MathAbstracts)mean exposure: 1.4 -> 8.1Hurt
reduce smoothing parameter m from 10 to 1 (ablation)SCP-Delta_r · MathAbstracts validationnext-token accuracy: 0.495 -> 0.39Hurt
increase smoothing parameter m from 10 to 1000 (ablation)SCP-Delta_r · MathAbstracts validationnext-token accuracy: 0.495No clear effect
increase smoothing parameter m (ablation)SCP-Delta_r · Canary (MathAbstracts)protectionHurt
insert canary 1 time versus 3 times (ablation)CP-Delta_r · Canary (MathAbstracts)mean exposure: 2.6 -> 3.6No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy