Metraipolis › archive › dc_b3fa0b05e7be

When Protection Overrides Consequences: Understanding Protection Mode in Large Language Model Decision-Making — arXiv:2610.19948

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.19948
License (as archived)
CC0-1.0
Archived
2026-10-15T20:02:13.092Z
SHA-256 of the archived text
d4a66c038ef906aa48d8482cdcc9a5ada0e5bc3a56bfa90ac1c0d83eb200a1cd
Stable URL of this page
https://metraipolis.com/documento/dc_b3fa0b05e7be

How to cite

When Protection Overrides Consequences: Understanding Protection Mode in Large Language Model Decision-Making — arXiv:2610.19948 — arXiv (en), https://arxiv.org/abs/2610.19948 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b3fa0b05e7be (Metraipolis archive dc_b3fa0b05e7be, sha256 d4a66c038ef906aa48d8482cdcc9a5ada0e5bc3a56bfa90ac1c0d83eb200a1cd)
BibTeX
@misc{dcb3fa0b05e7be,
  title = {When Protection Overrides Consequences: Understanding Protection Mode in Large Language Model Decision-Making — arXiv:2610.19948},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.19948},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b3fa0b05e7be; sha256 d4a66c038ef906aa48d8482cdcc9a5ada0e5bc3a56bfa90ac1c0d83eb200a1cd},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.19948. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Frontier language models excel at complex tasks yet may prioritize protection over larger physical consequences in simple, urgent decisions. GPT-5.6 Sol, for example, moves a girl playing in shallow water before retrieving a submerged camera. We call this Protection Mode and study it with ProtectionBench: 285 prompts, 34 physical families, and nine models. At lowest effort, at least one of four responses prioritizes the subject on 34.74% of Sol prompts and over 83% of MiniMax M3 and Qwen3.5-27B prompts; none of four independent human raters does so. Controls show that choices usually track physical risk. We elicit both risks before requesting an action. Some models assess risk incorrectly; others state the correct ordering but still choose the subject. Qwen3.5-27B with thinking shows 100% correct risk ordering but 57.81% subject choice among eligible conversations on eight selected scenes. Attention audits find no stable reduction in attention to relevant facts; removing the stated risk direction reduces object-first choice in dots3-note-prev but not detectably in Qwen3.5-27B. Adding a general "people or animals before property" rule increases subject choice by 19-38 percentage points, whereas factual reminders favor the object by 0.7-5.2 points. Removing naturally generated versions of that rule lowers subject choice by 10-15 points. We also examine whether post-training introduces this pattern. Base models already show Protection Mode; post-training changes justifications more than choices.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.19948 — 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 2026-10-15T19:35:48.279Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).