CheckMIABench: Firm Foundations For Membership Inference Attacks on Language Models
On the map: LLM safety and security → Membership inference attacks
Abstract (arXiv)
Membership inference attacks (MIAs) are a canonical way to assess a machine learning model's privacy properties. Although several attempts have been made to evaluate MIAs on language models, the extant literature has suffered numerous difficulties in constructing clean evaluations to test new techniques. In particular, subtle distribution shifts between member and non-member sets can undermine the statistical validity of MIAs; recent work has underscored this by showing that "blind" methods with no access to the underlying model can perform far better than published methods on the same benchmarks. This paper constructs a benchmark for principled evaluation of MIAs against LLMs, by leveraging the insight that training data before and after a fixed point during training are drawn from the same distribution. Therefore, all open-source models with intermediate checkpoints and public training data can be converted into MIA testbeds. We apply our framework to a half-dozen published attacks on the Pythia and OLMo family of models, from 70M to 7B parameters. To facilitate further privacy research, we open-source a modular library for designing and implementing attacks in this setting: https://github.com/safr-ai-lab/pandora_llm.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Placed in Membership inference attacks, one of 12 directions in LLM safety and security. Builds on 17 of its 22 references that are in the map. First posted on arXiv 16 Jun 2026.
Builds on
17 of 22 references.
Closest ancestors · 6
Same direction, newest first
- 2024Detecting Training Data of Large Language Models via Expectation Maximization10 Oct
- 2024SoK: Membership Inference Attacks on LLMs are Rushing Nowhere (and How to Fix It)25 Jun
- 2024Blind Baselines Beat Membership Inference Attacks for Foundation Models23 Jun
- 2024ReCaLL: Membership Inference via Relative Conditional Log-Likelihoods23 Jun
- 2023Membership Inference Attacks against Language Models via Neighbourhood Comparison29 May
- 2021Membership Inference Attacks From First Principles7 Dec
Other references · 11
- 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
- 2023LLaMA: Open and Efficient Foundation Language Modelscited by 14,132
- 2013Efficient Estimation of Word Representations in Vector Spacecited by 4,375
- 2020The Pile: An 800GB Dataset of Diverse Text for Language Modelingcited by 2,411
- 2016Membership Inference Attacks against Machine Learning Modelscited by 2,281
- 2020Extracting Training Data from Large Language Modelscited by 2,189
- 2023Pythia: A Suite for Analyzing Large Language Models Across Training and Scalingcited by 1,786
- 2023In-Context Unlearning: Language Models as Few Shot Unlearnerscited by 224
- 2024Inexact Unlearning Needs More Careful Evaluations to Avoid a False Sense of Privacycited by 82
- 2023MoPe: Model Perturbation-based Privacy Attacks on Language Modelscited by 25
- 2024Probing Language Models for Pre-training Data Detectioncited by 25
Built on by
Citing papers per year: 2026: 3.
3 papers.
Similar, not cited
Close in wording, no citation either way
- 2025Membership Inference Attacks on Large-Scale Models: A Survey
- 2025Membership Inference Attacks as Privacy Tools: Reliability, Disparity and Ensemble
- 2024Scaling Up Membership Inference: When and How Attacks Succeed on Large Language Models
- 2024Do Membership Inference Attacks Work on Large Language Models?
- 2026What Hard Tokens Reveal: Exploiting Low-confidence Tokens for Membership Inference Attacks against Large Language Models
- 2026A Full-Pipeline Framework for Evaluating Membership Inference Attacks in Machine Learning
- 2024EM-MIAs: Enhancing Membership Inference Attacks in Large Language Models through Ensemble Modeling
- 2026Powerful Training-Free Membership Inference Against Autoregressive Language Models