Metraipolis › archive › dc_bd930c5545b2

ASPIRE: Agentic Safety & Prompt Injection Red-teaming Engine — arXiv:2610.08951

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.08951
License (as archived): CC0-1.0
Archived: 2026-10-08T11:44:20.885Z
SHA-256 of the archived text: 2681b7a556e192109ba8ec5adb698764495b889b92f48e8a39029732472db2be
Stable URL of this page: https://metraipolis.com/documento/dc_bd930c5545b2

How to cite

ASPIRE: Agentic Safety & Prompt Injection Red-teaming Engine — arXiv:2610.08951 — arXiv (en), https://arxiv.org/abs/2610.08951 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bd930c5545b2 (Metraipolis archive dc_bd930c5545b2, sha256 2681b7a556e192109ba8ec5adb698764495b889b92f48e8a39029732472db2be)

BibTeX

@misc{dcbd930c5545b2,
  title = {ASPIRE: Agentic Safety & Prompt Injection Red-teaming Engine — arXiv:2610.08951},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.08951},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bd930c5545b2; sha256 2681b7a556e192109ba8ec5adb698764495b889b92f48e8a39029732472db2be},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.08951. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

LLM agents retrieve untrusted content and act through tools, creating indirect prompt-injection risks that can cause unauthorized actions or persistent state changes. Existing automated red-teaming largely optimizes payloads for pre-specified scenarios, leaving latent vulnerabilities across the agent's behavior space unexplored. We present ASPIRE, an Agentic Safety & Prompt Injection Red-teaming Engine for open-ended, behavior-level vulnerability discovery. ASPIRE maintains an evolving Agent Security Behavior Graph and uses complementary Explore and Exploit experts to discover, verify, and generalize consequence-centric tests. Trajectory evidence updates the graph and diagnoses partial or failed attempts, while cross-run memory transfers useful red-team strategies. Experiments on various benchmarks show that ASPIRE substantially expands coverage across consequences, injection methods, environments, and behavior paths while maintaining strong attack success.

Tema degli esploratori: fiducia digitale e crittografia (elenco arXiv cs.CR, nuovi lavori del giorno).
Categorie arXiv: Cryptography and Security (cs.CR).
Pagina del lavoro: https://arxiv.org/abs/2610.08951 — 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 2026-10-08T10:45:22.435Z dalla pagina https://arxiv.org/list/cs.CR/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).