mlsanityData to 14 Oct 2026

UnPII: Unlearning Personally Identifiable Information with Quantifiable Exposure Risk

arXiv 2601.01786 ↗First posted 5 Jan 2026cs.LG cs.CRcited by 0

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

The ever-increasing adoption of Large Language Models in critical sectors like finance, healthcare, and government raises privacy concerns regarding the handling of sensitive Personally Identifiable Information (PII) during training. In response, regulations such as European Union's General Data Protection Regulation (GDPR) mandate the deletion of PII upon requests, underscoring the need for reliable and cost-effective data removal solutions. Machine unlearning has emerged as a promising direction for selectively forgetting data points. However, existing unlearning techniques typically apply a uniform forgetting strategy that neither accounts for the varying privacy risks posed by different PII attributes nor reflects associated business risks. In this work, we propose UnPII, the first PII-centric unlearning approach that prioritizes forgetting based on the risk of individual or combined PII attributes. To this end, we introduce the PII risk index (PRI), a composite metric that incorporates multiple dimensions of risk factors: identifiability, sensitivity, usability, linkability, permanency, exposability, and compliancy. The PRI enables a nuanced evaluation of privacy risks associated with PII exposures and can be tailored to align with organizational privacy policies. To support realistic assessment, we systematically construct a synthetic PII dataset (e.g., 1,700 PII instances) that simulates realistic exposure scenarios. UnPII seamlessly integrates with established unlearning algorithms, such as Gradient Ascent, Negative Preference Optimization, and Direct Preference Optimization, without modifying their underlying principles. Our experimental results demonstrate that UnPII achieves the improvements of accuracy up to 11.8%, utility up to 6.3%, and generalizability up to 12.4%, respectively, while incurring a modest fine-tuning overhead of 27.5% on average during unlearning.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 19 of its 40 references that are in the map. First posted on arXiv 5 Jan 2026.

Builds on

19 of 40 references.

Closest ancestors · 8

Same direction, newest first

  1. 2024Unlearning as multi-task optimization: A normalized gradient difference approach with an adaptive learning rate29 Oct
  2. 2024MUSE: Machine Unlearning Six-Way Evaluation for Language Models8 Jul
  3. 2024Unveiling Entity-Level Unlearning for Large Language Models: A Comprehensive Analysis22 Jun
  4. 2024Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning8 Apr
  5. 2024TOFU: A Task of Fictitious Unlearning for LLMs11 Jan
  6. 2023FairSISA: Ensemble Post-Processing to Improve Fairness of Unlearning in LLMs12 Dec
  7. 2023Unlearn What You Want to Forget: Efficient Unlearning for LLMs31 Oct
  8. 2023Separate the Wheat from the Chaff: Model Deficiency Unlearning via Parameter-Efficient Module Operation16 Aug

Other references · 11

  1. 2017Attention Is All You Needcited by 60,361
  2. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  3. 2019Parameter-Efficient Transfer Learning for NLPcited by 4,236
  4. 2019Machine Unlearningcited by 1,061
  5. 2020Approximate Data Deletion from Machine Learning Modelscited by 255
  6. 2022Quantifying Privacy Risks of Masked Language Models Using Membership Inference Attackscited by 175
  7. 2023Beyond Memorization: Violating Privacy Via Inference with Large Language Modelscited by 175
  8. 2023ProPILE: Probing Privacy Leakage in Large Language Modelscited by 129
  9. 2024AutoTutor meets Large Language Models: A Language Model Tutor with Rich Pedagogy and Guardrailscited by 31
  10. 2024Evaluating LLM-based Personal Information Extraction and Countermeasurescited by 12
  11. 2021Detection of Illicit Drug Trafficking Events on Instagram: A Deep Multimodal Multilabel Learning Approachcited by 9

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Towards Benchmarking Privacy Vulnerabilities in Selective Forgetting with Large Language Models
  2. 2026Shadow Unlearning: A Neuro-Semantic Approach to Fidelity-Preserving Faceless Forgetting in LLMs
  3. 2025Rectifying Privacy and Efficacy Measurements in Machine Unlearning: A New Inference Attack Perspective
  4. 2025Efficient Unlearning with Privacy Guarantees
  5. 2025Towards Reliable Forgetting: A Survey on Machine Unlearning Verification
  6. 2026Mitigating Privacy Risk via Forget Set-Free Unlearning
  7. 2026Unintended Memorization of Sensitive Information in Fine-Tuned Language Models
  8. 2026Auditing Machine Unlearning: A Systematic Research on Whether Models Truly Forget