GAMBIT: A Gamified Jailbreak Framework for Multimodal Large Language Models

TL;DR

GAMBIT is a novel multimodal jailbreak framework that uses gamified scenes and reasoning chains to effectively induce large language models to generate harmful content, revealing safety vulnerabilities.

Contribution

The paper introduces GAMBIT, a new multimodal jailbreak method that leverages structured reasoning and gamification to significantly increase attack success rates on LLMs.

Findings

GAMBIT achieves over 92% attack success rate on Gemini 2.5 Flash.

It outperforms existing baselines in inducing harmful outputs.

The framework demonstrates effectiveness on multiple reasoning and non-reasoning models.

Abstract

Multimodal Large Language Models (MLLMs) have become widely deployed, yet their safety alignment remains fragile under adversarial inputs. Previous work has shown that increasing inference steps can disrupt safety mechanisms and lead MLLMs to generate attacker-desired harmful content. However, most existing attacks focus on increasing the complexity of the modified visual task itself and do not explicitly leverage the model's own reasoning incentives. This leads to them underperforming on reasoning models (Models with Chain-of-Thoughts) compared to non-reasoning ones (Models without Chain-of-Thoughts). If a model can think like a human, can we influence its cognitive-stage decisions so that it proactively completes a jailbreak? To validate this idea, we propose GAMBI} (Gamified Adversarial Multimodal Breakout via Instructional Traps), a novel multimodal jailbreak framework that decomposes and reassembles harmful visual semantics, then constructs a gamified scene that drives the model to explore, reconstruct intent, and answer as part of winning the game. The resulting structured reasoning chain increases task complexity in both vision and text, positioning the model as a participant whose goal pursuit reduces safety attention and induces it to answer the reconstructed malicious query. Extensive experiments on popular reasoning and non-reasoning MLLMs demonstrate that GAMBIT achieves high Attack Success Rates (ASR), reaching 92.13% on Gemini 2.5 Flash, 91.20% on QvQ-MAX, and 85.87% on GPT-4o, significantly outperforming baselines.