Multi-turn jailbreak attacks demonstrate that harmful intent can be distributed across dialogue, yet existing methods obscure what conversational mechanisms drive vulnerability. We introduce Blueprint, a safety-evaluation framework that separates a factorized social-influence strategy space from WorldviewSim, a cross-turn situational context module. Monte Carlo Tree Search optimizes turn-level combinations of 18 theory-grounded influence factors across a four-turn trajectory. Across six frontier models, Blueprint achieves near-ceiling attack success rates on major open-weight and proprietary models, while requiring the fewest average queries (2.46). The resulting trajectories further reveal model-specific vulnerability among resistant targets: each responds to distinct influence factors and strategy transitions, yet all share a common recovery pathway. Specifically, shifting toward concrete, executable task framing consistently escapes hard-refusal states. Ablations confirm that operational cues matter most: making requests actionable has the largest impact, gain framing is unusually potent, and some legitimacy appeals can backfire. These findings suggest that robust multi-turn safety requires monitoring not only harmful content, but also how dialogue state makes unsafe requests appear concrete and locally executable. The code is open-sourced at: https://github.com/charlottec1583/Blueprint.

Large Language Models (LLMs), despite advanced general capabilities, still suffer from numerous safety risks, especially jailbreak attacks that bypass safety protocols. Understanding these vulnerabilities through black-box jailbreak attacks, which better reflect real-world scenarios, offers critical insights into model robustness. While existing methods have shown improvements through various prompt engineering techniques, their success remains limited against safety-aligned models, overlooking a more fundamental problem: the effectiveness is inherently bounded by the predefined strategy spaces. However, expanding this space presents significant challenges in both systematically capturing essential attack patterns and efficiently navigating the increased complexity. To better explore the potential of expanding the strategy space, we address these challenges through a novel framework that decomposes jailbreak strategies into essential components based on the Elaboration Likelihood Model (ELM) theory and develops genetic-based optimization with intention evaluation mechanisms. To be striking, our experiments reveal unprecedented jailbreak capabilities by expanding the strategy space: we achieve over 90% success rate on Claude-3.5 where prior methods completely fail, while demonstrating strong cross-model transferability and surpassing specialized safeguard models in evaluation accuracy. The code is open-sourced at: https://github.com/Aries-iai/CL-GSO.

The keyphrase generation task is a challenging work that aims to generate a set of keyphrases for a piece of text. Many previous studies based on the sequence-to-sequence model were used to generate keyphrases, and they introduce a copy mechanism to achieve good results. However, we observed that most of the keyphrases are composed of some important words (seed words) in the source text, and if these words can be identified accurately and copied to create more keyphrases, the performance of the model might be improved. To address this challenge, we propose a DualCopyNet model, which introduces an additional sequence labeling layer for identifying seed words, and further copies the words for generating new keyphrases by dual copy mechanisms. Experimental results demonstrate that our model outperforms the baseline models and achieves an obvious performance improvement.