Challenge: Current approaches to address this issue involve costly dataset scrubbing or model filtering through unlearning and model editing.
Approach: They propose a method for unlearning sensitive information from language models . they curate email and URL datasets and a social security number dataset .
Outcome: The proposed method shows superior performance and robustness to extraction attacks on real-world datasets.

Similar Papers

Knowledge Unlearning for Mitigating Privacy Risks in Language Models (2023.acl-long)

Copied to clipboard

Challenge: Recent work shows that an adversary can extract training data from Pretrained Language Models including Personally Identifiable Information (PII) such as names, phone numbers, and email addresses.
Approach: They propose to use knowledge unlearning to reduce privacy risks for LMs by performing gradient ascent on target token sequences instead of trying to unlearn all the data at once.
Outcome: The proposed method can give a stronger empirical privacy guarantee in scenarios where the data vulnerable to extraction attacks are known a priori while being much more efficient and robust.
Unintended Memorization of Sensitive Information in Fine-Tuned Language Models (2026.eacl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) on sensitive datasets carry a substantial risk of unintended memorization and leakage of Personally Identifiable Information (PII) prior studies have analyzed memorizing dynamics in LLMs during pre-training and fine-tuning.
Approach: They investigate the vulnerability of PII that appears only in model inputs, not in training targets.
Outcome: The proposed methods show that post-training methods provide more consistent privacy-utility trade-offs .
DEPN: Detecting and Editing Privacy Neurons in Pretrained Language Models (2023.emnlp-main)

Copied to clipboard

Challenge: Existing studies have demonstrated that pretrained language models memorize and regurgitate a significant portion of training data, including atypical data points that appear only once in the training data.
Approach: They propose a method to locate and erase risky neurons in order to eliminate the impact of privacy data in the model in batches.
Outcome: The proposed method eliminates the impact of privacy data in the model in batches without affecting the model's performance.
Adapting Language Models When Training on Privacy-Transformed Data (2022.lrec-1)

Copied to clipboard

Challenge: Using data sanitization methods to remove personal information from spoken messages is not effective because privacy-transformed data is unlikely to match the test distribution.
Approach: They propose to use a data sanitization approach to remove personal information from spoken messages by replacing named entities with other words from the same class.
Outcome: The proposed approach removes personal information from the spoken messages using an automatic named entity recognition method.
Exploiting the Shadows: Unveiling Privacy Leaks through Lower-Ranked Tokens in Large Language Models (2025.acl-long)

Copied to clipboard

Challenge: Large language models face vulnerabilities related to the extraction of sensitive information.
Approach: They propose a method to exploit the model's lower-ranked output tokens to extract private information from retrieved documents or training knowledge.
Outcome: The proposed method is effective in both the agentic application privacy extraction setting and the direct training data extraction.
R.R.: Unveiling LLM Training Privacy through Recollection and Ranking (2025.findings-acl)

Copied to clipboard

Challenge: Existing privacy attacks focus on membership inference or data extraction, but reconstructing specific personally identifiable information (PII) in training data remains challenging.
Approach: They propose a two-step privacy stealing attack that enables attackers to reconstruct PII entities from scrubbed training data where the PI I entities have been masked.
Outcome: The proposed attack can reconstruct PII entities from scrubbed training data where the PI I entities have been masked.
PATS: Sensitivity-aware Noisy Learning for Pretrained Language Models (2022.emnlp-main)

Copied to clipboard

Challenge: A wide range of NLP tasks benefit from fine-tuning of pretrained language models (PLMs), however, a number of redundant parameters which contribute less to the downstream task are observed in a directly fine- tuned model.
Approach: They propose a noisy training mechanism which considers each parameter’s importance in the downstream task to help fine-tune pretrained language models.
Outcome: The proposed method can be used to fine-tune pretrained language models on a wide range of tasks and consistently achieve higher performance.
Tokens for Learning, Tokens for Unlearning: Mitigating Membership Inference Attacks in Large Language Models via Dual-Purpose Training (2025.findings-acl)

Copied to clipboard

Challenge: Existing defenses for large language models do not account for the sequential nature of text data.
Approach: They propose a lightweight yet effective empirical privacy defense that leverages token-specific characteristics to protect training data of large language models.
Outcome: The proposed approach provides strong protection against membership inference attacks and improves language modeling performance by 10% across different LLM architectures and datasets compared to baselines.
Preserving Privacy Through Dememorization: An Unlearning Technique For Mitigating Memorization Risks In Language Models (2023.emnlp-main)

Copied to clipboard

Challenge: Large Language models (LLMs) are trained on vast amounts of data, including sensitive information that poses a risk to personal privacy if exposed.
Approach: They propose a novel unlearning approach that utilizes an efficient reinforcement learning feedback loop via proximal policy optimization to incentivize the LLMs to learn a paraphrasing policy to unlearn the pre-training data.
Outcome: The proposed approach surpasses strong baselines and state-of-the-art methods in terms of its ability to generalize and strike a balance between privacy and LLM performance.
Fine-grained Pluggable Gradient Ascent for Knowledge Unlearning in Language Models (2024.emnlp-main)

Copied to clipboard

Challenge: Existing studies on knowledge unlearning focus on computer vision but extend their exploration to other fields.
Approach: They propose an adaptive objective that calculates gradients with fine-grained control specifically targeting sensitive tokens.
Outcome: The proposed method improves the general ability of language models while achieving knowledge unlearning.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations