Metraipolis › archive › dc_e3151dd52d86

AdvSim2Real : Training Web Agents Against Adaptive Prompt Injection in a Web World Model — arXiv:2610.08773

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.08773
License (as archived)
CC0-1.0
Archived
2026-10-07T11:22:10.428Z
SHA-256 of the archived text
de2a18579f88c7fdc84ac01e0aa22475ddb7d657b731c2486fd6874e48e01db1
Stable URL of this page
https://metraipolis.com/documento/dc_e3151dd52d86

How to cite

AdvSim2Real : Training Web Agents Against Adaptive Prompt Injection in a Web World Model — arXiv:2610.08773 — arXiv (en), https://arxiv.org/abs/2610.08773 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e3151dd52d86 (Metraipolis archive dc_e3151dd52d86, sha256 de2a18579f88c7fdc84ac01e0aa22475ddb7d657b731c2486fd6874e48e01db1)
BibTeX
@misc{dce3151dd52d86,
  title = {AdvSim2Real : Training Web Agents Against Adaptive Prompt Injection in a Web World Model — arXiv:2610.08773},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.08773},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e3151dd52d86; sha256 de2a18579f88c7fdc84ac01e0aa22475ddb7d657b731c2486fd6874e48e01db1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.08773. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Web agents complete user requests by reading and acting on pages that third parties write, so an instruction planted on a page can redirect the agent away from the user's goal. The agent cannot simply ignore the page, because the page also holds the values and controls the task requires. Current defenses fine-tune the agent on injections fixed before training, and attackers that adapt to the trained model bypass them. Adversarial training lets the attacker adapt but keeps the tasks fixed, so a task stops teaching once the agent solves it. We introduce AdvSim2Real, which co-evolves a task curriculum, an injection adversary, and the agent inside a frozen web world model. The curriculum is rewarded for tasks the agent solves about half of the time, and the adversary only for a success flip, an injection that turns a judged success into a failure. Training in the simulator makes a 4B agent both more capable and more robust: its completion rises with and without attacks, holds against a frontier-model adversary it never trained against, and its capability gain carries over to a real browser. On 150 web tasks, AdvSim2Real raises completion under this unseen adversary by 33.6\% relative to the base agent.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.08773 — 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 2026-10-07T11:15:24.865Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).