Machine Unlearning for Large Language Models: Foundations, Advances, and Agentic Extensions
cs.CR
Submitted: 2026-09-25
Updated: 2026-09-25
Comments: Under submission
Code: https://github.com/mlfoundations/evalchemy
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Open Problems in Machine Unlearning for AI Safety
- Dated Data: Tracing Knowledge Cutoffs in Large Language Models
- Inference-time Unlearning Using Conformal Prediction
- Who's Harry Potter? Approximate Unlearning in LLMs
- LLM Unlearning Under the Microscope: A Full-Stack View on Methods and Metrics
- Retrieval-Augmented Generation for Large Language Models: A Survey
- Stable Forgetting: Bounded Parameter-Efficient Unlearning in Foundation Models
- A Comprehensive Survey of Machine Unlearning Techniques for Large Language Models
- MEOW: MEMOry Supervised LLM Unlearning Via Inverted Facts
- FSFM: A Biologically-Inspired Framework for Selective Forgetting of Agent Memory
- FairSISA: Ensemble Post-Processing to Improve Fairness of Unlearning in LLMs
- Selective Forgetting for Large Reasoning Models
- Deployment-Time Memorization in Foundation-Model Agents
- FUNU: Boosting Machine Unlearning Efficiency by Filtering Unnecessary Unlearning
- AgentEHR: Advancing Autonomous Clinical Decision-Making via Retrospective Summarization
- Eight Methods to Evaluate Robust Unlearning in LLMs
- A Survey on Unlearning in Large Language Models
- SoK: Machine Unlearning for Large Language Models
- Spurious Rewards: Rethinking Training Signals in RLVR
- Step-by-Step Reasoning Attack: Revealing 'Erased' Knowledge in Large Language Models
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs