Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2511.00203 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866911244174229504 |
|---|---|
| author | Lüdke, David Wollschläger, Tom Ungermann, Paul Günnemann, Stephan Schwinn, Leo |
| author_facet | Lüdke, David Wollschläger, Tom Ungermann, Paul Günnemann, Stephan Schwinn, Leo |
| contents | We introduce a novel framework that transforms the resource-intensive (adversarial) prompt optimization problem into an \emph{efficient, amortized inference task}. Our core insight is that pretrained, non-autoregressive generative LLMs, such as Diffusion LLMs, which model the joint distribution over prompt-response pairs, can serve as powerful surrogates for prompt search. This approach enables the direct conditional generation of prompts, effectively replacing costly, per-instance discrete optimization with a small number of parallelizable samples. We provide a probabilistic analysis demonstrating that under mild fidelity assumptions, only a few conditional samples are required to recover high-reward (harmful) prompts. Empirically, we find that the generated prompts are low-perplexity, diverse jailbreaks that exhibit strong transferability to a wide range of black-box target models, including robustly trained and proprietary LLMs. Beyond adversarial prompting, our framework opens new directions for red teaming, automated prompt optimization, and leveraging emerging Flow- and Diffusion-based LLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_00203 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Diffusion LLMs are Natural Adversaries for any LLM Lüdke, David Wollschläger, Tom Ungermann, Paul Günnemann, Stephan Schwinn, Leo Machine Learning We introduce a novel framework that transforms the resource-intensive (adversarial) prompt optimization problem into an \emph{efficient, amortized inference task}. Our core insight is that pretrained, non-autoregressive generative LLMs, such as Diffusion LLMs, which model the joint distribution over prompt-response pairs, can serve as powerful surrogates for prompt search. This approach enables the direct conditional generation of prompts, effectively replacing costly, per-instance discrete optimization with a small number of parallelizable samples. We provide a probabilistic analysis demonstrating that under mild fidelity assumptions, only a few conditional samples are required to recover high-reward (harmful) prompts. Empirically, we find that the generated prompts are low-perplexity, diverse jailbreaks that exhibit strong transferability to a wide range of black-box target models, including robustly trained and proprietary LLMs. Beyond adversarial prompting, our framework opens new directions for red teaming, automated prompt optimization, and leveraging emerging Flow- and Diffusion-based LLMs. |
| title | Diffusion LLMs are Natural Adversaries for any LLM |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2511.00203 |