mlsanityData to 14 Oct 2026

Stronger Re-identification Attacks through Reasoning and Aggregation

arXiv 2510.09184 ↗First posted 10 Oct 2025cs.CLcited by 0

Not on the map. Why?

Abstract (arXiv)

Text de-identification techniques are often used to mask personally identifiable information (PII) from documents. Their ability to conceal the identity of the individuals mentioned in a text is, however, hard to measure. Recent work has shown how the robustness of de-identification methods could be assessed by attempting the reverse process of _re-identification_, based on an automated adversary using its background knowledge to uncover the PIIs that have been masked. This paper presents two complementary strategies to build stronger re-identification attacks. We first show that (1) the _order_ in which the PII spans are re-identified matters, and that aggregating predictions across multiple orderings leads to improved results. We also find that (2) reasoning models can boost the re-identification performance, especially when the adversary is assumed to have access to extensive background knowledge.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 14 of its 26 references that are in the map. First posted on arXiv 10 Oct 2025.

Builds on

14 of 26 references.

References · 14

  1. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  2. 2020REALM: Retrieval-Augmented Language Model Pre-Trainingcited by 2,141
  3. 2020ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERTcited by 1,178
  4. 2023In-Context Retrieval-Augmented Language Modelscited by 702
  5. 2024Smarter, Better, Faster, Longer: A Modern Bidirectional Encoder for Fast, Memory Efficient, and Long Context Finetuning and Inferencecited by 554
  6. 2023DeID-GPT: Zero-shot Medical Text De-Identification by GPT-4cited by 130
  7. 2022The Text Anonymization Benchmark (TAB): A Dedicated Corpus and Evaluation Framework for Text Anonymizationcited by 95
  8. 2016De-identification of Patient Notes with Recurrent Neural Networkscited by 88
  9. 2025Tower+: Bridging Generality and Translation Specialization in Multilingual LLMscited by 40
  10. 2014C-sanitized: a privacy model for document redaction and sanitizationcited by 27
  11. 2021No Intruder, no Validity: Evaluation Criteria for Privacy-Preserving Text Anonymizationcited by 13
  12. 2023De-identification of clinical free text using natural language processing: A systematic review of current approachescited by 13
2 more
  1. 2024BMX: Entropy-weighted Similarity and Semantic-enhanced Lexical Searchcited by 7
  2. 2025Re-identification of De-identified Documents with Autoregressive Infillingcited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026From Weak Cues to Real Identities: Evaluating Inference-Driven De-Anonymization in LLM Agents
  2. 2026AgentDoxx: Agentic Re-identification of Anonymized Text with Web Search
  3. 2025R.R.: Unveiling LLM Training Privacy through Recollection and Ranking
  4. 2026LLM Anonymization Against Agentic Re-Identification
  5. 2024Reconstructing training data from document understanding models
  6. 2022Similarity Distribution based Membership Inference Attack on Person Re-identification
  7. 2026SoK: Reconstruction Attacks on Synthetic Tabular Data (Insights from Winning the NIST CRC)
  8. 2026Backdoor in the Loop: Compromising Agentic Search via Malicious Retrievers