Metraipolis › archive › dc_a433824cdf44
Bit-Flip Attacks on Vision-Language-Action Models: Action-Decoding Architecture Shapes the Vulnerability — arXiv:2608.15475
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.15475
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T18:03:32.427Z
- SHA-256 of the archived text
98f5b1e80bef4a48fe143a1fcd1130c1717e7eb5d0cb05bce98a53252d14cd96- Stable URL of this page
https://metraipolis.com/documento/dc_a433824cdf44
How to cite
Bit-Flip Attacks on Vision-Language-Action Models: Action-Decoding Architecture Shapes the Vulnerability — arXiv:2608.15475 — arXiv (en), https://arxiv.org/abs/2608.15475 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a433824cdf44 (Metraipolis archive dc_a433824cdf44, sha256 98f5b1e80bef4a48fe143a1fcd1130c1717e7eb5d0cb05bce98a53252d14cd96)
BibTeX
@misc{dca433824cdf44,
title = {Bit-Flip Attacks on Vision-Language-Action Models: Action-Decoding Architecture Shapes the Vulnerability — arXiv:2608.15475},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.15475},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a433824cdf44; sha256 98f5b1e80bef4a48fe143a1fcd1130c1717e7eb5d0cb05bce98a53252d14cd96},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.15475. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Quantized Vision-Language-Action (VLA) models expose a weight-fault surface: Rowhammer-style faults can corrupt deployed INT8 bits. We present the first bit-flip attack on a VLA: a few gradient-selected flips reduce closed-loop success to $0\%$, while hundreds of random flips are harmless. Across four model variants spanning three action-head families, damaging bits concentrate in a few action-generating layers, but the empirical budget depends sharply on the head: direct regression and token policies fall in $1$--$5$ flips, whereas the evaluated flow-matching policies require ${\sim}100$--$300$. Our fixed-direction manifold-escape loss cuts \pizero{}'s budget from ${\sim}1000$ to ${\sim}100$ flips, and a matched five-direction sweep shows that the attack is not specific to an all-positive direction. On a direct head, protecting $3.1\%$ of weights preserves $60\%$ success at $K{=}100$, and protecting $5.3\%$ moves the open-loop break threshold from 3 to 100 flips. Finally, task-calibrated emulated $K{=}100$ flips yield $0/20$ real-robot successes, versus $14/20$ clean and $16/20$ global-random. Weight integrity is therefore a security boundary for embodied foundation models. Code is included as ancillary material.
Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.AI).
Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.15475 — 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 2026-10-18T16:29:14.022Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).