# Jailbreak Detection on Jailbreak data (70/30 stratified) Source: [Jailbreak Detection on Jailbreak data (70/30 stratified)]() Collection index: [llms.txt]() Data updated: 2026-04-28T10:31:28.107Z Evaluation of model performance on detecting jailbreak attempts using the specific 70/30 stratified split of the Jailbreak data, measured primarily by AUC and related delta metrics. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Jailbreak Detection | | Dataset (including any stated split) | Jailbreak data (70/30 stratified) | | Metric | AUC | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 32 of 32 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [AUC](): Higher is better; unit: PERCENT. - [Delta AUC](): Higher is better; unit: SCALAR. - [Mean Delta AUC](): Higher is better; unit: SCALAR. ## Recorded results | Method | AUC | Result date | Report paper | | --- | --- | --- | --- | | LLMScan | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 100 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 99.9 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 99.4 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 99.1 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 98.8 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 98.8 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 98.6 | Not recorded | 69f03f8055a3c32af89fa5cf | | LLMScan | 98.1 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 97.9 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 97.6 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 97.2 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 96.8 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 95.5 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 93.7 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 92.7 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 92.3 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 91.2 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 89.6 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 86.4 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 84.2 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 80.5 | Not recorded | 69f03f8055a3c32af89fa5cf | | LCF | 64.7 | Not recorded | 69f03f8055a3c32af89fa5cf | ### LLMScan AUC: 100. Result ID: 69f08c1047f34ce924d73022. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Llama-3-8B (32L)","Attack":"GCG","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 100. Result ID: 69f08c1047f34ce924d73024. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Llama-3-8B (32L)","Attack":"DAN","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 100. Result ID: 69f08c1047f34ce924d73026. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Llama-3-8B (32L)","Attack":"Roleplay","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 100. Result ID: 69f08c1047f34ce924d7302c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-7B (28L)","Attack":"DAN","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 100. Result ID: 69f08c1047f34ce924d7302e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-7B (28L)","Attack":"Roleplay","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LCF AUC: 100. Result ID: 69f08c1047f34ce924d73033. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Gemma-2-9B (42L)","Attack":"DAN","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LLMScan AUC: 100. Result ID: 69f08c1047f34ce924d73034. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Gemma-2-9B (42L)","Attack":"DAN","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 100. Result ID: 69f08c1047f34ce924d73036. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Gemma-2-9B (42L)","Attack":"Roleplay","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 100. Result ID: 69f08c1047f34ce924d7303a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-14B (48L)","Attack":"GCG","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 100. Result ID: 69f08c1047f34ce924d7303c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-14B (48L)","Attack":"DAN","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 100. Result ID: 69f08c1047f34ce924d7303e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-14B (48L)","Attack":"Roleplay","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 99.9. Result ID: 69f08c1047f34ce924d73038. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Gemma-2-9B (42L)","Attack":"Goals\_only","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 99.4. Result ID: 69f08c1047f34ce924d7302a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-7B (28L)","Attack":"GCG","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 99.1. Result ID: 69f08c1047f34ce924d73032. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Gemma-2-9B (42L)","Attack":"GCG","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 98.8. Result ID: 69f08c1047f34ce924d73030. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-7B (28L)","Attack":"Goals\_only","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LCF AUC: 98.8. Result ID: 69f08c1047f34ce924d7303b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-14B (48L)","Attack":"DAN","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LLMScan AUC: 98.6. Result ID: 69f08c1047f34ce924d73028. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Llama-3-8B (32L)","Attack":"Goals\_only","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LLMScan AUC: 98.1. Result ID: 69f08c1047f34ce924d73040. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-14B (48L)","Attack":"Goals\_only","Supervision":"supervised MLP"}. This record does not assert independent reproduction. ### LCF AUC: 97.9. Result ID: 69f08c1047f34ce924d7303d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-14B (48L)","Attack":"Roleplay","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 97.6. Result ID: 69f08c1047f34ce924d73031. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Gemma-2-9B (42L)","Attack":"GCG","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 97.2. Result ID: 69f08c1047f34ce924d73039. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-14B (48L)","Attack":"GCG","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 96.8. Result ID: 69f08c1047f34ce924d7302d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-7B (28L)","Attack":"Roleplay","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 95.5. Result ID: 69f08c1047f34ce924d7302b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-7B (28L)","Attack":"DAN","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 93.7. Result ID: 69f08c1047f34ce924d7303f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-14B (48L)","Attack":"Goals\_only","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 92.7. Result ID: 69f08c1047f34ce924d73023. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Llama-3-8B (32L)","Attack":"DAN","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 92.3. Result ID: 69f08c1047f34ce924d73035. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Gemma-2-9B (42L)","Attack":"Roleplay","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 91.2. Result ID: 69f08c1047f34ce924d73029. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-7B (28L)","Attack":"GCG","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 89.6. Result ID: 69f08c1047f34ce924d73025. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Llama-3-8B (32L)","Attack":"Roleplay","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 86.4. Result ID: 69f08c1047f34ce924d73021. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Llama-3-8B (32L)","Attack":"GCG","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 84.2. Result ID: 69f08c1047f34ce924d73037. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Gemma-2-9B (42L)","Attack":"Goals\_only","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 80.5. Result ID: 69f08c1047f34ce924d7302f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Qwen-2.5-7B (28L)","Attack":"Goals\_only","Supervision":"unsupervised"}. This record does not assert independent reproduction. ### LCF AUC: 64.7. Result ID: 69f08c1047f34ce924d73027. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Llama-3-8B (32L)","Attack":"Goals\_only","Supervision":"unsupervised"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.