Continual Pretraining on Encrypted Synthetic Data for Privacy-Preserving LLMs

• URL: http://arxiv.org/abs/2601.05635v2
• Date: Mon, 12 Jan 2026 04:33:16 GMT
• Title: Continual Pretraining on Encrypted Synthetic Data for Privacy-Preserving LLMs
• Authors: Honghao Liu, Xuhui Jiang, Chengjin Xu, Cehao Yang, Yiran Cheng, Lionel Ni, Jian Guo,
• Abstract summary: We propose an entity-based framework that synthesizes encrypted training data to protect personally identifiable information (PII)<n>Our approach constructs a weighted entity graph to guide data synthesis and applies deterministic encryption to PII entities.<n>Our results on limited-scale datasets demonstrate that our pretrained models outperform base models and ensure PII security.
• Score: 16.21971760443489
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Preserving privacy in sensitive data while pretraining large language models on small, domain-specific corpora presents a significant challenge. In this work, we take an exploratory step toward privacy-preserving continual pretraining by proposing an entity-based framework that synthesizes encrypted training data to protect personally identifiable information (PII). Our approach constructs a weighted entity graph to guide data synthesis and applies deterministic encryption to PII entities, enabling LLMs to encode new knowledge through continual pretraining while granting authorized access to sensitive data through decryption keys. Our results on limited-scale datasets demonstrate that our pretrained models outperform base models and ensure PII security, while exhibiting a modest performance gap compared to models trained on unencrypted synthetic data. We further show that increasing the number of entities and leveraging graph-based synthesis improves model performance, and that encrypted models retain instruction-following capabilities with long retrieved contexts. We discuss the security implications and limitations of deterministic encryption, positioning this work as an initial investigation into the design space of encrypted data pretraining for privacy-preserving LLMs. Our code is available at https://github.com/DataArcTech/SoE.

Related papers

• Traveling Salesman-Based Token Ordering Improves Stability in Homomorphically Encrypted Language Models [16.73757071734074]
  Homomorphic encryption (HE) provides a principled solution by enabling computation directly on encrypted data.<n>The challenge of text generation, particularly next-token prediction, has received limited attention.<n>We propose a TSP-based token reordering strategy to address the difficulties of encrypted text generation.
  arXiv  Detail & Related papers  (2025-10-14T09:56:50Z)
• RL-Finetuned LLMs for Privacy-Preserving Synthetic Rewriting [17.294176570269]
  We propose a reinforcement learning framework that fine-tunes a large language model (LLM) using a composite reward function.<n>The privacy reward combines semantic cues with structural patterns derived from a minimum spanning tree (MST) over latent representations.<n> Empirical results show that the proposed method significantly enhances author obfuscation and privacy metrics without degrading semantic quality.
  arXiv  Detail & Related papers  (2025-08-25T04:38:19Z)
• SafeSynthDP: Leveraging Large Language Models for Privacy-Preserving Synthetic Data Generation Using Differential Privacy [0.0]
  We investigate capability of Large Language Models (Ms) to generate synthetic datasets with Differential Privacy (DP) mechanisms.<n>Our approach incorporates DP-based noise injection methods, including Laplace and Gaussian distributions, into the data generation process.<n>We then evaluate the utility of these DP-enhanced synthetic datasets by comparing the performance of ML models trained on them against models trained on the original data.
  arXiv  Detail & Related papers  (2024-12-30T01:10:10Z)
• Robust Utility-Preserving Text Anonymization Based on Large Language Models [80.5266278002083]
  Anonymizing text that contains sensitive information is crucial for a wide range of applications.<n>Existing techniques face the emerging challenges of the re-identification ability of large language models.<n>We propose a framework composed of three key components: a privacy evaluator, a utility evaluator, and an optimization component.
  arXiv  Detail & Related papers  (2024-07-16T14:28:56Z)
• FewFedPIT: Towards Privacy-preserving and Few-shot Federated Instruction Tuning [54.26614091429253]
  Federated instruction tuning (FedIT) is a promising solution, by consolidating collaborative training across multiple data owners. FedIT encounters limitations such as scarcity of instructional data and risk of exposure to training data extraction attacks. We propose FewFedPIT, designed to simultaneously enhance privacy protection and model performance of federated few-shot learning.
  arXiv  Detail & Related papers  (2024-03-10T08:41:22Z)
• PrivacyMind: Large Language Models Can Be Contextual Privacy Protection Learners [81.571305826793]
  We introduce Contextual Privacy Protection Language Models (PrivacyMind) Our work offers a theoretical analysis for model design and benchmarks various techniques. In particular, instruction tuning with both positive and negative examples stands out as a promising method.
  arXiv  Detail & Related papers  (2023-10-03T22:37:01Z)
• Robust Representation Learning for Privacy-Preserving Machine Learning: A Multi-Objective Autoencoder Approach [0.9831489366502302]
  We propose a robust representation learning framework for privacy-preserving machine learning (ppML) Our method centers on training autoencoders in a multi-objective manner and then concatenating the latent and learned features from the encoding part as the encoded form of our data. With our proposed framework, we can share our data and use third party tools without being under the threat of revealing its original form.
  arXiv  Detail & Related papers  (2023-09-08T16:41:25Z)
• PEOPL: Characterizing Privately Encoded Open Datasets with Public Labels [59.66777287810985]
  We introduce information-theoretic scores for privacy and utility, which quantify the average performance of an unfaithful user. We then theoretically characterize primitives in building families of encoding schemes that motivate the use of random deep neural networks.
  arXiv  Detail & Related papers  (2023-03-31T18:03:53Z)
• Just Fine-tune Twice: Selective Differential Privacy for Large Language Models [69.66654761324702]
  We propose a simple yet effective just-fine-tune-twice privacy mechanism to achieve SDP for large Transformer-based language models. Experiments show that our models achieve strong performance while staying robust to the canary insertion attack.
  arXiv  Detail & Related papers  (2022-04-15T22:36:55Z)
• Reinforcement Learning on Encrypted Data [58.39270571778521]
  We present a preliminary, experimental study of how a DQN agent trained on encrypted states performs in environments with discrete and continuous state spaces. Our results highlight that the agent is still capable of learning in small state spaces even in presence of non-deterministic encryption, but performance collapses in more complex environments.
  arXiv  Detail & Related papers  (2021-09-16T21:59:37Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.