A Cross-Architecture Audit of Direction-Based Inference-Time Defences in Vision-Language Models
Inference time defences against vision language model jailbreaks often subtract a calibrated direction from the residual stream at a chosen decoder layer. We compare five defence candidates across 15 model and layer cells from four architectural families under a magnitude controlled protocol that matches the intervention size for each prompt and pairs every direction with a random control of the same norm. The candidates are the mean image conditioning shift, a CMRM style refusal direction, a ShiftDC style attack specific residual, a prompt instruction to ignore the image, and a random control. No single candidate dominates on both refusal recovery and utility preservation. The image cond...
| Authors | Xiangyu Yin, Tora Bodin, Rohan Menon et al. |
|---|---|
| Primary category | cs.AI |
| Published | 30 July 2026 |
| Last updated at source | 23 August 2026 |
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.