RefusalGuard: Geometry-Preserving Fine-Tuning for Safety in LLMs
Fine-tuning safety-aligned language models for downstream tasks often leads to substantial degradation of refusal behavior, making models vulnerable to adversarial misuse. While prior work has shown that safety-relevant features are encoded in structured representations within the model's activation space, how these representations change during fine-tuning and why alignment degrades remains poorly understood. In this work, we investigate the representation-level mechanisms underlying alignment degradation. Our analysis shows that standard fine-tuning induces systematic drift in safety-relevant representations, distorts their geometric structure, and introduces interference between task o...
| Authors | Sadia Asif, Mohammad Mohammadi Amiri |
|---|---|
| Primary category | cs.LG |
| Published | 3 May 2026 |
| Last updated at source | 21 August 2026 |
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.