PapersSelected

GUARD: Natural Forgetting in Large Reasoning Models

GUARD: Natural Forgetting in Large Reasoning Models via Guided Answer-Reasoning Distillation

Why we picked it

This research provides a method to make large language models forget sensitive information, which is important for privacy and safety in AI applications.

A new method called GUARD addresses the challenge of unlearning in large reasoning models by creating a natural forgetting trajectory.

Body

It converts unsafe disclosures into safe, non-disclosing reasoning paths and refines the model's behavior through distillation.

Experiments on R-TOFU and a STAR-1-derived setting show GUARD reduces unsafe disclosures while preserving reasoning utility.

Original · arXiv cs.AI

GUARD: Natural Forgetting in Large Reasoning Models via Guided Answer-Reasoning Distillation

Recent advances in large reasoning models (LRMs) have made machine unlearning more challenging, as protected facts or unsafe rationales may surface in intermediate chain-of-thought (CoT) traces before the final answer is produced. Existing unlearning objectives typically suppress the target content or redirect internal representations, but they never specify how the post-forgetting trajectory should continue, which can lead to hallucinated substitutes, malformed boundaries, or repetitive outputs. We argue that LRM unlearning should instead learn a natural forgetting trajectory: a coherent non-disclosing CoT followed by a stable refusal-style answer that replace the original disclosure. To this end, we propose Guided Answer-Reasoning Distillation (GUARD), which converts model-generated unsafe disclosures into safe-exit trajectories, aligns a frozen LRM via guidance tokens, and distills the guided behavior into model parameters.To address the lack of metrics for replacement quality beyond leakage, we further introduce Natural Forgetting Reasoning Score (NFRS), which captures structural stability, fluency, and unsupported substitutes in forgotten outputs. Extensive experiments on R-TOFU and a STAR-1-derived harmful-intent setting show that GUARD substantially reduces unsafe and privacy disclosures across two widely adopted distilled LRMs while preserving reasoning utility. Codes are available at https://github.com/zeyu-Yan/GUARD

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include GUARD, large language models, reasoning models.

Source attribution: curated from arXiv cs.AI (@Zeyu Yan, Guanghao Zhou, Minghui Qiu, Ming Gao, Cen Chen).