Metraipolis › archive › dc_cd8cb20d8443
TACS: Trajectory-Aware Candidate Selection for LLM Jailbreak Suffix Optimization — arXiv:2608.29564
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.29564
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T08:07:01.880Z
- SHA-256 of the archived text
1463e48232d98dc32f88d9b71048cf2b6aea24814cded37d00e7f7d07ca91a9e- Stable URL of this page
https://metraipolis.com/documento/dc_cd8cb20d8443
How to cite
TACS: Trajectory-Aware Candidate Selection for LLM Jailbreak Suffix Optimization — arXiv:2608.29564 — arXiv (en), https://arxiv.org/abs/2608.29564 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cd8cb20d8443 (Metraipolis archive dc_cd8cb20d8443, sha256 1463e48232d98dc32f88d9b71048cf2b6aea24814cded37d00e7f7d07ca91a9e)
BibTeX
@misc{dccd8cb20d8443,
title = {TACS: Trajectory-Aware Candidate Selection for LLM Jailbreak Suffix Optimization — arXiv:2608.29564},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.29564},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cd8cb20d8443; sha256 1463e48232d98dc32f88d9b71048cf2b6aea24814cded37d00e7f7d07ca91a9e},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.29564. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Gradient-based jailbreak suffix optimization methods typically update the suffix by retaining the candidate with the lowest current loss. We show that this seemingly natural design is fundamentally myopic: candidates that look better under the current-step proxy often fail to produce better jailbreak outcomes later in the search, revealing a form of selection-stage reward hacking. This suggests that candidate selection, rather than candidate generation alone, is a hidden bottleneck in suffix optimization. To address this issue, we propose TACS, a trajectory-aware candidate selection framework for jailbreak suffix optimization. Instead of selecting candidates solely by their immediate loss, TACS augments per-step evaluation with a trajectory-aware proxy and stabilizes selection with reference-policy regularization and a discriminator-estimated chi-squared correction, encouraging choices that remain effective beyond the current step. Experiments on HarmBench show that TACS consistently outperforms strong baselines under the same search budget, substantially improving attack success rates while exhibiting more stable optimization behavior throughout the search. Our findings highlight that mitigating selection-stage reward hacking caused by myopic candidate selection is critical for improving jailbreak suffix optimization. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.CL). Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2608.29564 — 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 2026-10-17T07:47:52.709Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).