mlsanityData to 14 Oct 2026

AlignDP: Hybrid Differential Privacy with Rarity-Aware Protection for LLMs

arXiv 2512.17251 ↗First posted 19 Dec 2025cs.CR cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Large language models are exposed to risks of extraction, distillation, and unauthorized fine-tuning. Existing defenses use watermarking or monitoring, but these act after leakage. We design AlignDP, a hybrid privacy lock that blocks knowledge transfer at the data interface. The key idea is to separate rare and non-rare fields. Rare fields are shielded by PAC indistinguishability, giving effective zero-epsilon local DP. Non-rare fields are privatized with RAPPOR, giving unbiased frequency estimates under local DP. A global aggregator enforces composition and budget. This two-tier design hides rare events and adds controlled noise to frequent events. We prove limits of PAC extension to global aggregation, give bounds for RAPPOR estimates, and analyze utility trade-off. A toy simulation confirms feasibility: rare categories remain hidden, frequent categories are recovered with small error.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 8 of its 9 references that are in the map. First posted on arXiv 19 Dec 2025.

Builds on

8 of 9 references.

References · 8

  1. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  2. 2016Deep Learning with Differential Privacycited by 2,728
  3. 2023Extracting Training Data from Diffusion Modelscited by 732
  4. 2023A Watermark for Large Language Modelscited by 725
  5. 2016Stealing Machine Learning Models via Prediction APIscited by 619
  6. 2014RAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal Responsecited by 272
  7. 2019High Accuracy and High Fidelity Extraction of Neural Networkscited by 214
  8. 2024Large Language Model Unlearning via Embedding-Corrupted Promptscited by 116

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Benchmarking Empirical Privacy Protection for Adaptations of Large Language Models
  2. 2024Privacy-Preserving Retrieval-Augmented Generation with Differential Privacy
  3. 2026Securing LLMs in the Wild: Privacy and Security Challenges at the Edge
  4. 2026DP-OPD: Differentially Private On-Policy Distillation for Language Models
  5. 2026DPrivBench: Benchmarking LLMs' Reasoning for Differential Privacy
  6. 2025DP-Fusion: Token-Level Differentially Private Inference for Large Language Models
  7. 2024Privacy-Preserving Large Language Models: Mechanisms, Applications, and Future Directions
  8. 2026Private Prediction via PAC Privacy