arXiv Computation and Language September 2, 2026 By Shengfang Zhai, Leo Marchyok, Yuling Shi, Huanran Chen, Yinpeng Dong, Jiaheng Zhang, Sanghyun Hong

Membership Inference in Fine-tuned Diffusion Language Models via Token-level Memorization Asymmetry

Read the original on arXiv Computation and Language →

The paper introduces a new privacy vulnerability in diffusion language models (DLMs) called token‑level memorization asymmetry, derived from theoretical analysis of diffusion training dynamics. It proposes Q‑Skew, a quantile‑weighted skewness indicator, to perform membership inference on fine‑tuned DLMs, outperforming existing baselines across multiple datasets and models. Additionally, Q‑Skew can be used to extract personally identifiable information (PII), demonstrating a broader privacy attack surface.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.

arXiv AI

1d ago

Batch Normalization Amplifies Memorization and Privacy Risks

Batch Normalization (BN) is widely used to speed up and stabilize deep neural network training, yet its effect on privacy and memorization has been largely unexplored. This study shows that BN significantly increases the memorization of atypical or outlier samples, as evidenced by unintended memorization, per-sample influence, and heightened susceptibility to membership inference attacks across multiple datasets and architectures. A mechanistic analysis of the BN backward pass reveals that BN amplifies the per‑step margin growth of outlier samples during training, thereby intensifying their influence.

By Ngoc Phu Doan, Chongyan Gu, Ihsen Alouani

safety

arXiv Machine Learning

Sep 11

Adaptive Diffusion Freezing: Privacy-preserving Diffusion Models Against Membership Inference Attacks

Adaptive Diffusion Freezing (ADF) is a new privacy‑preserving framework for diffusion models that protects against membership inference attacks (MIAs). It uses cross‑timestep adaptive freezing training, where a mask matrix controls which data subsets participate at each diffusion timestep, reducing over‑memorization and aligning model behavior for member and non‑member samples. A pretraining‑based risk‑aware freezing policy estimates MIA risk and suppresses high‑risk subset‑timestep pairs, achieving a superior privacy‑utility‑efficiency trade‑off across multiple datasets.

By Jialu Guo, Xiao Han, Junjie Wu

diffusion benchmarks safety

arXiv Machine Learning

Sep 1

Balancing Privacy, Utility, and Safety in LLM Alignment through Preference Optimization

arXiv:2608.30141v1 Announce Type: cross Abstract: Preference optimization is widely used to align large language models with human preferences, but preference-data composition may also influence priv...

By Dishu Yang, Jingjing Liu, Jize Li

llms reinforcement-learning efficiency safety

arXiv Machine Learning

Jul 21

Gradient-Free Privacy Leakage in Federated Language Models through Selective Weight Tampering

arXiv:2310. 16152v5 Announce Type: replace-cross Abstract: Federated learning (FL) has become a key component in various language modeling applications such as machine translation, next-word prediction, and medical record analysis.

By Md Rafi Ur Rashid, Vishnu Asutosh Dasu, Kang Gu, Najrin Sultana, Shagufta Mehnaz

llms nlp safety

arXiv Machine Learning

Jun 2

Causal Evaluation of Membership Inference Attacks

arXiv:2602. 02819v4 Announce Type: replace Abstract: Membership Inference Attacks (MIAs) aim to distinguish training points (members) from unseen data (non-members), and are widely used to quantify memorization and assess privacy risks.

By Mathieu Even, Cl\'ement Berenfeld, Linus Bleistein, Tudor Cebere, Julie Josse, Aur\'elien Bellet

llms fine-tuning safety

arXiv Machine Learning

Jun 17

CheckMIABench: Firm Foundations For Membership Inference Attacks on Language Models

arXiv:2606. 17464v1 Announce Type: new Abstract: Membership inference attacks (MIAs) are a canonical way to assess a machine learning model's privacy properties.

By Jeffrey G. Wang, Jason Wang, Marvin Li, Seth Neel

llms benchmarks safety

One email a morning, machine-written

Related stories

Batch Normalization Amplifies Memorization and Privacy Risks

Adaptive Diffusion Freezing: Privacy-preserving Diffusion Models Against Membership Inference Attacks

Balancing Privacy, Utility, and Safety in LLM Alignment through Preference Optimization

Gradient-Free Privacy Leakage in Federated Language Models through Selective Weight Tampering

Causal Evaluation of Membership Inference Attacks

CheckMIABench: Firm Foundations For Membership Inference Attacks on Language Models