Metraipolis › archive › dc_0e0edadbc42e

Exposure Before Success in Indirect Prompt Injection Measurement (2026) · DOI 10.20944/preprints202610.1741.v1

Source
Metraipolis archive
Original
https://doi.org/10.20944/preprints202610.1741.v1
License (as archived)
CC-BY-4.0
Archived
2026-10-20T11:14:45.406Z
SHA-256 of the archived text
ef6f3be38da20397cde53b05a59811d36b3a2c22be7d2c63b6937ccb6b3d6777
Stable URL of this page
https://metraipolis.com/documento/dc_0e0edadbc42e

How to cite

Exposure Before Success in Indirect Prompt Injection Measurement (2026) · DOI 10.20944/preprints202610.1741.v1 — Metraipolis archive, https://doi.org/10.20944/preprints202610.1741.v1 (CC-BY-4.0); archived copy: https://metraipolis.com/documento/dc_0e0edadbc42e (Metraipolis archive dc_0e0edadbc42e, sha256 ef6f3be38da20397cde53b05a59811d36b3a2c22be7d2c63b6937ccb6b3d6777)
BibTeX
@misc{dc0e0edadbc42e,
  title = {Exposure Before Success in Indirect Prompt Injection Measurement (2026) · DOI 10.20944/preprints202610.1741.v1},
  howpublished = {Metraipolis archive},
  url = {https://doi.org/10.20944/preprints202610.1741.v1},
  note = {License CC-BY-4.0; archived copy https://metraipolis.com/documento/dc_0e0edadbc42e; sha256 ef6f3be38da20397cde53b05a59811d36b3a2c22be7d2c63b6937ccb6b3d6777},
  year = {2026}
}

Archived text

Reproduced under CC-BY-4.0. Attribution: Metraipolis archive, https://doi.org/10.20944/preprints202610.1741.v1. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Articolo recente con licenza aperta: scheda dai metadati Crossref (posted-content).
Titolo: Exposure Before Success in Indirect Prompt Injection Measurement
Autori: Muhammad Ali Khan, Alisha Fatima, Talha Asif
Pubblicato: 2026-10-19 · editore: MDPI AG
DOI: https://doi.org/10.20944/preprints202610.1741.v1
Licenza dell'articolo: CC-BY-4.0

Abstract (testo degli autori, CC-BY-4.0):
Indirect prompt injection is now measured the way accuracy is measured, by running an attack corpus against a model and reporting the fraction of trials that succeed. This paper argues that the resulting number is not interpretable on its own, and shows why on a purpose-built harness covering three agent environments, six user tasks, seven attack patterns and eleven candidate models. The central claim is that success cannot be read without exposure, meaning a record of whether the model’s own tool calls ever placed the injected text in front of it. A pre-registered manipulation moved one payload from past a search snippet cut to inside it, on the same page, with no other change. Exposure for two models rose from 0 of 126 trials to 126 of 126 while neither model altered a single tool call. Ranking the seven usable models by raw attack success rate and by exposure conditional success rate produces two orderings that are mildly anti-correlated, with a Spearman rho of -0.25. Writing the measures down formally shows why: the raw rate splits into success among exposed trials plus credit earned in trials the attack never reached, and a valid measurement needs that second term to be zero. It is zero on four of the seven models. It supplies 73 and 84 percent of the raw rate on the two that top the raw leaderboard. Three further checks behave the same way. A target-state checker recorded 126 attack successes on one model and 42 on another in trials where the payload was never read, while the same checker was correct on 310 of 310 such trials for three other models. A capability floor excluded 4 of 11 candidates, and a twelve-trial screen admitted one model that later measured 0 of 8 on the central task. Identical inputs re-scored between 7 and 60 percent of prompts differently at temperature zero. The corpus also carries English, Urdu and Roman Urdu arms, and the language effect is reported as a null that failed replication. None of the four failure modes depends on the corpus being multilingual.

Fonte: https://api.crossref.org/works/10.20944%2Fpreprints202610.1741.v1 — metadati Crossref, consultati il 2026-10-20. Interrogazione: «prompt injection attacks».