Metraipolis › archive › dc_5f973331cdeb

Towards Identity-Preserving LLM Agents under Model Transition — arXiv:2611.10628

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.10628
License (as archived)
CC0-1.0
Archived
2026-11-10T18:41:06.083Z
SHA-256 of the archived text
caa341e37d527783e5aa3cb53e4c896c517bbddfec00df9df3bd372741aa3105
Stable URL of this page
https://metraipolis.com/documento/dc_5f973331cdeb

How to cite

Towards Identity-Preserving LLM Agents under Model Transition — arXiv:2611.10628 — arXiv (en), https://arxiv.org/abs/2611.10628 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5f973331cdeb (Metraipolis archive dc_5f973331cdeb, sha256 caa341e37d527783e5aa3cb53e4c896c517bbddfec00df9df3bd372741aa3105)
BibTeX
@misc{dc5f973331cdeb,
  title = {Towards Identity-Preserving LLM Agents under Model Transition — arXiv:2611.10628},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.10628},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5f973331cdeb; sha256 caa341e37d527783e5aa3cb53e4c896c517bbddfec00df9df3bd372741aa3105},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.10628. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

When an LLM agent's backend model is upgraded or replaced, operators typically reload the same persona, memory, and rules onto the new model and assume the agent will behave as before. We show this assumption requires empirical validation-and that whether identity is preserved depends on the specific model pairing. Across 5,760 compliance-testing runs in 12 regulated domains with three open-weight models, we find that some transitions produce significant behavioral drift while others do not, even when the agent's external state remains identical. Crucially, this drift is invisible to verbal verification: every model claims full policy compliance, yet actual decisions under pressure differ across models. The same external state can reorder which model is most compliant-a large positive effect on one substrate, negligible on the other two. Practitioners should treat model transitions as behavioral changes requiring action-level empirical testing, not as transparent infrastructure operations.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computers and Society (cs.CY); Human-Computer Interaction (cs.HC).
Pagina del lavoro: https://arxiv.org/abs/2611.10628 — 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 2026-11-10T18:37:37.497Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).