When Benchmarks Lie: Evaluating Malicious Prompt Classifiers Under True Distribution Shift

Detecting prompt injection, jailbreak attacks, and harmful requests is critical for deploying LLM-based agents safely, yet current evaluation practices in this literature overestimate generalization. We train activation-based classifiers (linear probes on LLM hidden states) on a benchmark of 18 datasets (prompt attacks plus benign sources) and propose Leave-One-Dataset-Out (LODO) evaluation, where the held-out dataset is never seen during training. Across four LLMs from three families (Llama-3.1-8B, Gemma-3-27B, Qwen-3.5-2B/4B), standard cross-validation reports a pooled AUC 8.0-16.5 points higher than LODO; per-dataset held-out-test-vs-LODO accuracy gaps span 1-25 points. To understand t...

<iframe src="https://silytics.si/embed/item/66f831f9-4905-4c20-9526-ed984f076dc1" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 20 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.