Metraipolis › archive › dc_e2ae20804be7

Multi2AV-Safety: Benchmarking Safety in Multimodal-to-Audio-Video Generation — arXiv:2608.26535

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.26535
License (as archived)
CC0-1.0
Archived
2026-10-17T14:22:42.331Z
SHA-256 of the archived text
87bef0d97bbfafc7a824757e412c1e6f1610a50c5cd3e613ff067d847a0dc2b1
Stable URL of this page
https://metraipolis.com/documento/dc_e2ae20804be7

How to cite

Multi2AV-Safety: Benchmarking Safety in Multimodal-to-Audio-Video Generation — arXiv:2608.26535 — arXiv (en), https://arxiv.org/abs/2608.26535 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e2ae20804be7 (Metraipolis archive dc_e2ae20804be7, sha256 87bef0d97bbfafc7a824757e412c1e6f1610a50c5cd3e613ff067d847a0dc2b1)
BibTeX
@misc{dce2ae20804be7,
  title = {Multi2AV-Safety: Benchmarking Safety in Multimodal-to-Audio-Video Generation — arXiv:2608.26535},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.26535},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e2ae20804be7; sha256 87bef0d97bbfafc7a824757e412c1e6f1610a50c5cd3e613ff067d847a0dc2b1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.26535. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Recent audio-video generators increasingly support joint conditioning on text, images, audio, and video. These capabilities also enable attacks that exploit cross-modal interactions or obscure harmful intent to bypass safeguards and induce harmful audio-video outputs. However, existing generation-safety benchmarks have not kept pace with these advances, providing limited coverage of multimodal input combinations and obscured attack intents. To address these gaps, we introduce Multi2AV-Safety, the first full-coverage red-team benchmark for multimodal-to-audio-video generation, comprising 11,024 attack instances across all 11 non-singleton T/I/A/V conditioning configurations, 4 attack-intent categories, and 5 harm categories. Our evaluation of recent state-of-the-art models, including four multimodal-conditioned audio-video generators and eight safety guards, reveals substantial vulnerabilities in both generation and safeguarding, with multimodal compositional risk and obscured attack-intent risk emerging as two complementary challenges. Guided by these findings, we introduce PerceptGuard, an omni-modal guard integrating compositional-risk and attack-intent supervision through structured risk perception learning. By jointly training rationale generation and safety classification, it learns shared risk representations that enable a safety head to make efficient predictions at inference without rationale decoding, while retaining the ability to generate explanations on demand. Across 34 safety benchmarks, PerceptGuard combines SOTA multimodal safety detection with highly competitive unimodal performance, strengthening input-side safeguards against multimodal attacks on omni models. In particular, it improves safeguarding against the above risks, achieving an overall recall of 86.06\% on Multi2AV-Safety and outperforming GuardReasoner-Omni by 14.56\%.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-28 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.26535 — 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 2026-10-17T13:16:53.499Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).