Dual-Modality Multi-Stage Adversarial Safety Training: Robustifying Multimodal Web Agents Against Cross-Modal Attacks

Studies webpage injections that affect both screenshots and accessibility trees, then trains web agents through teacher imitation, supervised adversarial examples, and adversarial reinforcement learning. The evaluation tests defenses on MiniWoB++ and held-out tasks.

AI-assisted editorial review against a primary source. This is a catalog annotation, not an independent reproduction or endorsement.

What is this resource?

Resource type

Research papers

Environment

web

Topics

safety-security, reinforcement-learning, evaluation

Record publication date

2026-03-04

Evidence retrieved

2026-10-16T06:36:43Z

Resource ID

arxiv:2603.04364

Recorded paper version

arXiv v2

Version updated: 2026-10-05T23:01:49Z.

This identifies the retained arXiv metadata snapshot; newly observed changes require separate review.

Authors

Haoyu Liu, Dingcheng Li, Lukas Rutishauser, Zeyu Zheng

Cite this paper

Download BibTeX · Download CSL JSON

Citations preserve the selected source edition, its authors and recorded date precision. Unrecorded venues and identifiers are omitted.

Explore related topics

Scope and limitations

The annotation and scope were reviewed against the complete retained original arXiv abstract. Experimental results and artifact availability have not been independently reproduced.

Primary evidence and verification

Source retrieval is separate from experiment reproduction. Use the original publication or publisher page when citing claims about the resource.

1. arxiv api

   Retrieved 2026-10-16T06:36:43Z. Supports: name, authors, published_at, updated_at, identifiers.arxiv, details.arxiv_snapshot.

   Content fingerprint

   c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59
2. editorial primary review

   Retrieved 2026-10-16T06:36:43Z. Supports: name, kind, summary, environments, topics, scope, notes.

   Content fingerprint

   c432f419481a105a8ce71174ec4aea23c4f0d36ddf89cfe1d0974fa349f6cb59

Suggest a correction

Open a source-backed correction. Include this record ID and the original source. Read our editorial policy and coverage boundaries.