Operationalising the Right to be Forgotten in LLMs: A Lightweight Sequential Unlearning Framework for Privacy-Aligned Deployment in Politically Sensitive Environments

2026-04-14

AI Summary

This work addresses the challenge of deploying large language models (LLMs) in politically sensitive environments while complying with privacy regulations such as the GDPR’s “right to be forgotten.” The authors propose a lightweight sequential unlearning framework that first applies forward fine-tuning to stabilize the model’s general capabilities and then employs layer-constrained negative fine-tuning to precisely suppress sensitive content. By decoupling retention and forgetting objectives, this approach achieves low-overhead, reproducible privacy alignment. Experiments on the SemEval-2025 LLM unlearning benchmark demonstrate that the framework effectively eliminates sensitive behaviors without compromising factual accuracy or linguistic fluency. Notably, larger models (e.g., GPT-2 versus DistilGPT-2) exhibit stronger unlearning robustness, highlighting the critical role of model scale in privacy adaptation.

Technology Category

• Machine Learning: Privacy
• Natural Language Processing: (Large) Language Models
• Computer Vision: Large Vision Models

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Security and Privacy: Security and privacy of machine learning and AI applications

Abstract

Large Language Models (LLMs) are increasingly deployed in politically sensitive environments, where memorisation of personal data or confidential content raises regulatory concerns under frameworks such as the GDPR and its Right to be Forgotten. Translating such legal principles into large-scale generative systems presents significant technical challenges. We introduce a lightweight sequential unlearning framework that explicitly separates retention and suppression objectives. The method first stabilises benign capabilities through positive fine-tuning, then applies layer-restricted negative fine-tuning to suppress designated sensitive patterns while preserving general language competence. Experiments on the SemEval-2025 LLM Unlearning benchmark demonstrate effective behavioural suppression with minimal impact on factual accuracy and fluency. GPT-2 exhibits greater robustness than DistilGPT-2, highlighting the role of model capacity in privacy-aligned adaptation. We position sequential unlearning as a practical and reproducible mechanism for operationalising data erasure requirements in politically deployed LLMs.

Problem

• Right to be Forgotten
• Large Language Models
• Privacy
• Unlearning
• Politically Sensitive Environments

Innovation

• sequential unlearning
• right to be forgotten
• layer-restricted fine-tuning
• privacy-aligned LLMs
• negative fine-tuning