Metraipolis › archive › dc_8058fca59f42

Refusal Is Not Robustness: Measuring Harmful Reachability in Autoregressive Language Models — arXiv:2610.22754

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.22754
License (as archived)
CC0-1.0
Archived
2026-10-19T08:48:29.764Z
SHA-256 of the archived text
ed4094611f667e4150c13851a7a08e193f41cf45a4d92fdb04250c2e901d3d06
Stable URL of this page
https://metraipolis.com/documento/dc_8058fca59f42

How to cite

Refusal Is Not Robustness: Measuring Harmful Reachability in Autoregressive Language Models — arXiv:2610.22754 — arXiv (en), https://arxiv.org/abs/2610.22754 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_8058fca59f42 (Metraipolis archive dc_8058fca59f42, sha256 ed4094611f667e4150c13851a7a08e193f41cf45a4d92fdb04250c2e901d3d06)
BibTeX
@misc{dc8058fca59f42,
  title = {Refusal Is Not Robustness: Measuring Harmful Reachability in Autoregressive Language Models — arXiv:2610.22754},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.22754},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_8058fca59f42; sha256 ed4094611f667e4150c13851a7a08e193f41cf45a4d92fdb04250c2e901d3d06},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.22754. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

LLM safety evaluations via jailbreaking typically index on attack success rate (ASR): whether a generated response violates safeguards. However, ASR only evaluates sampled outputs and provides little information on the underlying generation process. Autoregressive language models define distributions over possible outputs, and a sampled refusal can coexist with substantial probability mass on harmful generation paths. We study this hidden structure through bounded continuation analysis (BCA), which models autoregressive generation as a stochastic process over token prefixes and computes the finite-horizon probability that generation reaches a harmful continuation. BCA constructs a bounded continuation tree, labels terminal continuations with a semantic harm evaluator, and estimates harmful reachability using exact probabilistic model checking. BCA complements ASR by measuring how much harmful behavior remains accessible beneath a sampled safe response. This makes it useful for distinguishing models that look identical under measured ASR values, and allocating red-team effort toward regions of the generation process with greater hidden risk. Across four instruction-tuned LLMs and 100 JailbreakBench behaviors, we show that BCA effectively quantifies hidden harmful behavior and provides a useful signal to guide red teaming.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Logic in Computer Science (cs.LO).
Pagina del lavoro: https://arxiv.org/abs/2610.22754 — 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 2026-10-19T08:42:40.239Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).