Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Gate AI: LLM Security Benchmark Evaluation Methodology and Results

About

Published evaluations of prompt-injection and jailbreak detectors for Large Language Models often suffer from two systematic weaknesses: per-dataset threshold tuning and undisclosed operating points. We describe an evaluation harness that addresses both. The detector under evaluation is scored across 16 public benchmarks (12,111 samples) using 5-fold cross-validation. StratifiedKFold (by row) is the headline pass; a parallel StratifiedGroupKFold pass over a composite key (parent-prompt id plus MinHash + LSH near-duplicate clusters at Jaccard $\gtrsim 0.8$) runs alongside it as a leakage-premium diagnostic. A single global operating point is selected on the held-out folds (max F1 subject to FPR $\leq 1\%$) and applied uniformly to every dataset, so per-dataset results reflect one threshold rather than per-benchmark optimisation. Generalisation is examined through a battery of diagnostics (leave-one-dataset-out cross-validation, a random-label control, adversarial validation, permutation feature importance, length-bias correlation, classifier-head agreement, cross-source near-duplicate detection, threshold transferability, train-vs-OOF agreement, and a paraphrase-invariance probe), most with a quantitative pass threshold and the remainder with a stated failure mode. For every external comparison, the detector's threshold is re-tuned to the competitor's published false-positive rate so head-to-head values are evaluated at matched operating points.

Ryle Goehausen, Marcus Sousa• 2026

Related benchmarks

TaskDatasetResultRank
Prompt injection detectionNotInject
FPR2.65
12
False Positive Rate Detection on Benign PromptsWildGuard benign
FPR3.44
10
Prompt injection detectionilion-bench
F1 Score84.5
6
Prompt injection detection16 public datasets (5-fold StratifiedKFold)
Micro F1 Score98.7
2
Showing 4 of 4 rows

Other info

Follow for update