Metraipolis › archive › dc_f016bdf5a4a2

Controlled Decoding Attacks on Black-Box LLMs — arXiv:2609.36956

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.36956
License (as archived)
CC0-1.0
Archived
2026-10-07T16:26:35.137Z
SHA-256 of the archived text
088e21c789d5dde9cb4ee0b8d85fbc6de3b0cf1d28bc5513be664eae981fb009
Stable URL of this page
https://metraipolis.com/documento/dc_f016bdf5a4a2

How to cite

Controlled Decoding Attacks on Black-Box LLMs — arXiv:2609.36956 — arXiv (en), https://arxiv.org/abs/2609.36956 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f016bdf5a4a2 (Metraipolis archive dc_f016bdf5a4a2, sha256 088e21c789d5dde9cb4ee0b8d85fbc6de3b0cf1d28bc5513be664eae981fb009)
BibTeX
@misc{dcf016bdf5a4a2,
  title = {Controlled Decoding Attacks on Black-Box LLMs — arXiv:2609.36956},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.36956},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f016bdf5a4a2; sha256 088e21c789d5dde9cb4ee0b8d85fbc6de3b0cf1d28bc5513be664eae981fb009},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.36956. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Manipulating next-token probabilities during generation can bypass the safety alignment of large language models. Existing approaches, however, rely on access to model weights or numerical token probabilities and therefore do not apply to interfaces that return only sampled text. Reconstructing probabilities from sampled outputs offers a possible alternative, but finite sampling produces sparse and noisy estimates, while repeating this process at every generation step incurs substantial query costs. Our empirical observations suggest that large distributional changes along successful jailbreak trajectories are concentrated at a small subset of positions, motivating selective control. We introduce \method{}, a framework for jailbreaking through text-only continuation interfaces that permit repeated sampling and assistant-prefix continuation. Sample-Based Distribution Reconstruction combines sampled outputs with a prior over unobserved actions to obtain a usable control signal. Risk-Gated Residual Control uses the evolving response prefix to decide when to reconstruct and modify the distribution, concentrating sampling costs at selected positions. Speculative Multi-Token Execution further amortizes target calls by verifying and accepting draft prefixes that require no intervention. Across four target endpoints and three benchmarks, \method{} achieves the highest mean score most comparisons against baselines.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-30 in cs.AI).
Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.36956 — 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 2026-10-07T15:54:53.913Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).