Metraipolis › archive › dc_192ac9ffe87a
On the Relative Security of Prompt Injection: Mechanised Definitions, Defences and Verdicts for LLM Agents — arXiv:2611.11506
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.11506
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-10T18:43:26.979Z
- SHA-256 of the archived text
eb7f75499375b8b6e776bf70e5ce6a0e6b0d9ce2bc2404b071a6200e9913b242- Stable URL of this page
https://metraipolis.com/documento/dc_192ac9ffe87a
How to cite
On the Relative Security of Prompt Injection: Mechanised Definitions, Defences and Verdicts for LLM Agents — arXiv:2611.11506 — arXiv (en), https://arxiv.org/abs/2611.11506 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_192ac9ffe87a (Metraipolis archive dc_192ac9ffe87a, sha256 eb7f75499375b8b6e776bf70e5ce6a0e6b0d9ce2bc2404b071a6200e9913b242)
BibTeX
@misc{dc192ac9ffe87a,
title = {On the Relative Security of Prompt Injection: Mechanised Definitions, Defences and Verdicts for LLM Agents — arXiv:2611.11506},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.11506},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_192ac9ffe87a; sha256 eb7f75499375b8b6e776bf70e5ce6a0e6b0d9ce2bc2404b071a6200e9913b242},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.11506. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
An LLM agent that reads untrusted content, such as an email or a web page, may follow instructions hidden in it. Noninterference is too strong for such agents, since a useful agent lets untrusted content influence what it does and may disclose the user's data by design, and declassification needs a release policy written for each task. We instead state security relative to a reference agent, which runs the same plan under the same runtime monitor with a model that never obeys untrusted content. An agent is relatively secure if every leak it has is a leak the reference already has. Our framework, PlanTERS, instantiates relative security, a notion from speculative execution, with a small agent language in which the attacker chooses how the model follows instructions, a composition with runtime monitors, and attacker models that differ in what they observe. A fence theorem shows that revoking injectable tools at the first untrusted read, and never granting them back, makes every plan relatively secure at the event level on finite runs, for every model behaviour in which each injection is a single call and the model obeys on a schedule fixed in advance; with a fairness condition, this extends to possibly infinite runs. We prove or refute relative security for agents with fences, budgets, approvals, loops and dual LLMs, in Isabelle/HOL. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2611.11506 — 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 2026-11-10T18:37:37.644Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).