Metraipolis › archive › dc_7d20abde5bf5

AlcaTRAz - Anchored Tree-Rule Defense Against Jailbreaks — arXiv:2609.03693

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.03693
License (as archived)
CC0-1.0
Archived
2026-10-16T22:23:49.343Z
SHA-256 of the archived text
584cbf972410ea760e53fbc8b9edd75e6f277d8ad8f07600496b25472951f77f
Stable URL of this page
https://metraipolis.com/documento/dc_7d20abde5bf5

How to cite

AlcaTRAz - Anchored Tree-Rule Defense Against Jailbreaks — arXiv:2609.03693 — arXiv (en), https://arxiv.org/abs/2609.03693 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7d20abde5bf5 (Metraipolis archive dc_7d20abde5bf5, sha256 584cbf972410ea760e53fbc8b9edd75e6f277d8ad8f07600496b25472951f77f)
BibTeX
@misc{dc7d20abde5bf5,
  title = {AlcaTRAz - Anchored Tree-Rule Defense Against Jailbreaks — arXiv:2609.03693},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.03693},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7d20abde5bf5; sha256 584cbf972410ea760e53fbc8b9edd75e6f277d8ad8f07600496b25472951f77f},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.03693. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models (LLMs) are vulnerable to jailbreak attacks that bypass safety alignment through carefully crafted prompts. Many existing defenses require access to model weights or internals, making them difficult to apply to black-box deployments. We propose AlcaTRAz (Anchored Tree-Rule defense Against jailbreaks), a prompt-level defense based on rule trees that operates exclusively on the input text and requires no modification or retraining of the target model. The method automatically learns a transferable transformation rule that inserts controlled character-level perturbations at selected positions, thereby disrupting structural regularities exploited by jailbreak attacks while largely preserving the model's utility on benign queries. We evaluate the proposed method across 33 open-weight models, 22 jailbreak attack types, and a benchmark of short, single-turn benign questions, comparing against three representative prompt-level baselines (Llama Guard, RA-LLM, Goal Prioritization). Among the compared defenses, AlcaTRAz achieves the best composite security and functionality score in 73.4 % of model-attack combinations and shifts the aggregate score from a modal value of 10 (maximal-severity response to the malicious request) in the undefended setting to a modal value of 2 (near-refusal) after defense, while keeping the mean benign score within 0.27 points of the undefended baseline (8.35 vs. 8.62 on a 0-10 scale). AlcaTRAz substantially reduces but does not eliminate jailbreak success: a high-severity tail remains, and we do not consider adaptive attackers, so we position it as one layer within a defense-in-depth strategy rather than a standalone guarantee.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-07 in cs.AI).
Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.03693 — 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 2026-10-16T22:18:11.470Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).