Metraipolis › archive › dc_2a9630f076dd
AnchorRep: Defending LLMs Against Cross-Model Adversarial Transfer via Representation Repulsion — arXiv:2609.32602
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.32602
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T21:00:27.104Z
- SHA-256 of the archived text
92419c0722d1c59dd519384d1626042a7c57d75b9b3f10e6271a407bf34c84dc- Stable URL of this page
https://metraipolis.com/documento/dc_2a9630f076dd
How to cite
AnchorRep: Defending LLMs Against Cross-Model Adversarial Transfer via Representation Repulsion — arXiv:2609.32602 — arXiv (en), https://arxiv.org/abs/2609.32602 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2a9630f076dd (Metraipolis archive dc_2a9630f076dd, sha256 92419c0722d1c59dd519384d1626042a7c57d75b9b3f10e6271a407bf34c84dc)
BibTeX
@misc{dc2a9630f076dd,
title = {AnchorRep: Defending LLMs Against Cross-Model Adversarial Transfer via Representation Repulsion — arXiv:2609.32602},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.32602},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2a9630f076dd; sha256 92419c0722d1c59dd519384d1626042a7c57d75b9b3f10e6271a407bf34c84dc},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32602. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Adversarial attacks optimized on a single open-weight LLM can transfer to and jailbreak architecturally different models, allowing an attacker with white-box access to one model to compromise independently deployed systems. This creates a shared vulnerability across models, yet existing defenses are not designed for this cross-model threat. We find that cross-model transfer aligns with shared internal representation geometry, making it a natural defense target. AnchorRep targets this geometry directly with a lightweight LoRA adapter that pushes the defended model's internal representations of harmful prompts away from those of a frozen anchor model on the same prompts. Training uses a small set of harmful prompts and no adversarial examples. Across five models and four architectural families, AnchorRep reduces cross-model attack success rate to <=1.1% on 2,000 transferred attacks (0% on two), including the largest drop on Mistral (36% -> 1.1%). Existing defenses can reduce transfer, but only at high cost either inducing up to 77% degenerate benign output or increasing over-refusal by up to 18%. Because such degenerate benign outputs are not captured by standard refusal-based metrics, we introduce the Benign Garble Rate to quantify them. Our results suggest that cross-model robustness can be achieved by shaping representation geometry, without requiring attack-specific training Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.32602 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T20:29:02.406Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).