NegBLEURT Forest: Leveraging Inconsistencies for Detecting Jailbreak Attacks
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914173913399296 |
|---|---|
| author | Sleem, Lama Francois, Jerome Li, Lujun Foucher, Nathan Gentile, Niccolo State, Radu |
| author_facet | Sleem, Lama Francois, Jerome Li, Lujun Foucher, Nathan Gentile, Niccolo State, Radu |
| contents | Jailbreak attacks designed to bypass safety mechanisms pose a serious threat by prompting LLMs to generate harmful or inappropriate content, despite alignment with ethical guidelines. Crafting universal filtering rules remains difficult due to their inherent dependence on specific contexts. To address these challenges without relying on threshold calibration or model fine-tuning, this work introduces a semantic consistency analysis between successful and unsuccessful responses, demonstrating that a negation-aware scoring approach captures meaningful patterns. Building on this insight, a novel detection framework called NegBLEURT Forest is proposed to evaluate the degree of alignment between outputs elicited by adversarial prompts and expected safe behaviors. It identifies anomalous responses using the Isolation Forest algorithm, enabling reliable jailbreak detection. Experimental results show that the proposed method consistently achieves top-tier performance, ranking first or second in accuracy across diverse models using the crafted dataset, while competing approaches exhibit notable sensitivity to model and data variations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_11784 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | NegBLEURT Forest: Leveraging Inconsistencies for Detecting Jailbreak Attacks Sleem, Lama Francois, Jerome Li, Lujun Foucher, Nathan Gentile, Niccolo State, Radu Cryptography and Security Artificial Intelligence Jailbreak attacks designed to bypass safety mechanisms pose a serious threat by prompting LLMs to generate harmful or inappropriate content, despite alignment with ethical guidelines. Crafting universal filtering rules remains difficult due to their inherent dependence on specific contexts. To address these challenges without relying on threshold calibration or model fine-tuning, this work introduces a semantic consistency analysis between successful and unsuccessful responses, demonstrating that a negation-aware scoring approach captures meaningful patterns. Building on this insight, a novel detection framework called NegBLEURT Forest is proposed to evaluate the degree of alignment between outputs elicited by adversarial prompts and expected safe behaviors. It identifies anomalous responses using the Isolation Forest algorithm, enabling reliable jailbreak detection. Experimental results show that the proposed method consistently achieves top-tier performance, ranking first or second in accuracy across diverse models using the crafted dataset, while competing approaches exhibit notable sensitivity to model and data variations. |
| title | NegBLEURT Forest: Leveraging Inconsistencies for Detecting Jailbreak Attacks |
| topic | Cryptography and Security Artificial Intelligence |
| url | https://arxiv.org/abs/2511.11784 |