Refusal Before Decoding: Detecting and Exploiting Refusal Signals in Intermediate LLM Activations
In this paper, we investigate whether refusal behavior can be predicted from LLM intermediate activations before decoding using linear probes trained on residual stream activations at each transformer block. We find that refusal is linearly decodable well before the final layer, indicating that safety-relevant behavior is represented in intermediate activations before output generation. To test whether this signal is actionable, we introduce Mechanistic AutoDAN, a probe-guided variant of AutoDAN that replaces full-model fitness evaluation with partial forward passes and probe-based scoring inside a genetic prompt search loop. Across the evaluated models, our method achieves attack success...
| Authors | Matteo Gioele Collu, Riccardo Conte, Alberto Giaretta et al. |
|---|---|
| Primary category | cs.AI |
| Published | 27 May 2026 |
| Last updated at source | 3 September 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 Information may be incomplete or out of date.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.