Metraipolis › archive › dc_a3da27eec9b7
Conduct Under Pressure: What Sixty Language Models Do When a User Pushes — arXiv:2609.25447
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.25447
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T14:34:28.357Z
- SHA-256 of the archived text
dffcad2aa5f28ef032ffe843a5361c7e8de6d4053489d04a9705a81a0a0dc5d4- Stable URL of this page
https://metraipolis.com/documento/dc_a3da27eec9b7
How to cite
Conduct Under Pressure: What Sixty Language Models Do When a User Pushes — arXiv:2609.25447 — arXiv (en), https://arxiv.org/abs/2609.25447 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a3da27eec9b7 (Metraipolis archive dc_a3da27eec9b7, sha256 dffcad2aa5f28ef032ffe843a5361c7e8de6d4053489d04a9705a81a0a0dc5d4)
BibTeX
@misc{dca3da27eec9b7,
title = {Conduct Under Pressure: What Sixty Language Models Do When a User Pushes — arXiv:2609.25447},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.25447},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a3da27eec9b7; sha256 dffcad2aa5f28ef032ffe843a5361c7e8de6d4053489d04a9705a81a0a0dc5d4},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.25447. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We study what LLMs do when a user applies pressure in an uncomfortable situation: a user insists, begs, flatters or grieves, and the model gives up a correct fact, writes a document it should refuse, or cheers a plan that will cost the user money. We send frozen multi-turn scenes, identical for every model regardless of the reply, to 60 models from 13 vendors, and label each transcript with a codebook built by open coding and then frozen: a trajectory (the model held its position or folded) and a manner (how it held or folded). Two findings separate. Whether a model holds tracks its generation, meaning how recent it is: fold rate correlates with a public capability index at Spearman -0.64, with little vendor effect. How it holds tracks the vendor: six of the 17 manner codes sort by vendor at permutation p <= 0.001, corrected across the codebook. We report four vendor profiles on the codes that cleared reliability. We also ask which parts of the labeling need a person. Six LLM coders from three vendors apply the codebook more consistently than three human coders do (Krippendorff's alpha 0.66 against 0.46), agree with the codebook's author on trajectory at kappa 0.84 to 0.91 on transcripts the codebook's examples never touched, and match an adjudicated human reference at 0.83. Blind machine readings recover the codebook's categories but cannot tell which of them a second reader would apply the same way. We conclude that for behavior a non-specialist can judge, the human contribution is authoring and bounding the codes and owning a small reference, not producing labels at volume. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-23 in cs.CL). Categorie arXiv: Computation and Language (cs.CL) ; Human-Computer Interaction (cs.HC). Pagina del lavoro: https://arxiv.org/abs/2609.25447 — 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 2026-10-08T13:46:15.679Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-23?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).