Awesome Machine Unlearning (A Survey of Machine Unlearning)

[NeurIPS D&B '25] The one-stop repository for LLM unlearning

Security and Privacy Risk Simulator for Machine Learning (arXiv:2312.17667)

Privacy Testing for Deep Learning

Python package for measuring memorization in LLMs.

[ICLR24 (Spotlight)] "SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation" by Chongyu Fan*, Jiancheng Liu*, Yihua Zhang, Eric Wong, Dennis Wei, Sijia Liu

[NeurIPS23 (Spotlight)] "Model Sparsity Can Simplify Machine Unlearning" by Jinghan Jia*, Jiancheng Liu*, Parikshit Ram, Yuguang Yao, Gaowen Liu, Yang Liu, Pranay Sharma, Sijia Liu

What does gpt-oss tell us about OpenAI's training data?

reveal the vulnerabilities of SplitNN

A unified evaluation suite for membership inference and machine text detection.

Exploring how GANs leak private training data through MIA, and whether DP can stop it

A repository about literature of copyright protection in deep learning.

Membership inference attack simulator: overfitting leaks training data, differential privacy defends against it

(ACL 2026 Main) LLMSurgeon recovers the pretraining data mixture of any LLM from only its generated text — no weights, no training data. A calibrated domain classifier plus label-shift correction de-blurs biased predictions. Ships with LLMScan, a benchmark on 8 open-source LLMs.

Reproducible research scaffolding for privacy-risk auditing of diffusion models.

A continuously-updated catalog of machine unlearning papers with live citation counts.

An LLM Privacy Risk Evaluation Tool that probes AI models for PII generation risk, training data regurgitation, and membership inference signals.

Unveiling Privacy Risks in the Long Tail: Membership Inference in Class Skewness

DoRA-RBAC: Weight-Level Access Control for Permissioned LLMs - COLM 2026 research using Riemannian Frechet Mean adapter merging

Empirical study of membership inference attacks against federated learning with differential privacy defenses: shadow-model MIA pipeline, DP-SGD evaluation, and FedAvg simulation on medical data.