Overview

Jailbreak attacks study how to make AI systems produce harmful or otherwise restricted outputs despite safety safeguards, using methods such as adversarial prompt optimization, encoded instructions, multi-turn persuasion, and search over attack strategies. Research now tests these methods not only on text-only language models but also on image, audio, and video systems, tool-using agents, and robots, with benchmarks including AdvBench, HarmBench, and JailbreakBench. Defenses include input classifiers, tracking intent across turns or tools, and interventions in model activations, but they must preserve useful responses to benign requests. A central challenge is that measured success depends on attack budgets, prompt variants, and evaluation judges, while defenses may fail when harmful intent is disguised by style, language, modality, or context.

Latest papers

79 papers6 papers in the last four weeks, down 45% on the four weeks before. 0.0% of all new papers.

Nov 5, 2026

"Blind Compliance": Coercive Control in AI Jailbreaking

The explosion of generative AI systems and concerns surrounding their rapidly scaling capabilities has led to an adversarial cycle between increasingly narrow content restrictions ("guardrails") and increasingly sophisticated structured prompting techniques ("jailbreaks"). Technological frameworks surrounding such jailbreaking remain largely focused on the system's output: whether harmful or disallowed content is ultimately produced for the user. However, this approach, which draws largely from classical cybersecurity, does not address the uniquely relational dimension introduced by generative AI: in many cases the user is not just accessing a technology via an exploit, but is in conversation with it as part of the exploit. Here we present an alternative lens which centers the interaction rather than the output. Drawing on interpersonal violence scholarship, we present a novel framework of coercive control categories with a particular focus on aspects which are embodied versus disembodied. We then map how these disembodied coercive control dynamics are structurally embedded in modern AI training paradigms, and how those embeddings are exploited via a particular suite of persona-based jailbreaks. In this lens, we reframe AI safety and alignment as not merely a technical security problem, but as another surface upon which we see enacted the relational dynamics of power and control.
Oct 29, 2026

CoFiBreak: A Coarse-to-Fine Diversity Reward for Effective Multi-Turn Jailbreak

Automated red teaming trains an attacker large language model (LLM) to elicit harmful responses from a victim LLM, but reinforcement learning can collapse attackers onto a narrow set of prompts when attack success is the only reward. Such a collapse of attack diversity can bias safety evaluation toward repeatedly testing near-duplicate attack patterns rather than the underlying harmful intents. Existing methods often trade off attack success or inference cost for greater content or strategy diversity. We propose CoFiBreak, a coarse-to-fine diversity reward for Group Relative Policy Optimization (GRPO) that improves attack diversity while preserving strong attack effectiveness and low inference cost. The coarse reward broadens which strategies the attacker uses by encouraging strategies underrepresented within each GRPO group, while the fine reward broadens how those strategies are realized by discouraging reuse of the same personas, framings, and output formats after harmful-intent information is removed from prompt embeddings. Diversity rewards apply only to attacks that already exceed an effectiveness threshold, so diversity cannot be gained through ineffective attacks. Since both rewards are computed only during training, CoFiBreak also retains inference efficiency. Trained with Llama3.1-8B-Instruct as the victim, CoFiBreak discovers 1.6-2.0x as many distinct successful fine-grained strategies as the strongest baseline across frontier victims on HarmBench. It reaches 98.8% ASR@10 (attack success rate with 10 generations per intent) on AdvBench with the highest diversity score among all compared methods, which is 2.1x that of the strongest-attacking baseline. The resulting attacks transfer to unseen frontier LLMs, reaching 71.0% ASR@10 on AdvBench with the unseen victim Claude-Opus-4.6. The code will be released.
Oct 28, 2026

Jailbreak Connectivity: Towards Diverse, Transferable, and Universal Multimodal Large Language Model Jailbreak

Multimodal large language models (MLLMs) are vulnerable to visual jailbreak attacks that bypass safety safeguards. Existing methods typically optimize isolated adversarial examples, limiting attack diversity, transferability, and generalization. We propose Jailbreak Connectivity (JC), a framework based on the observation that effective adversarial inputs can be connected through low-loss regions in image space. JC optimizes continuous adversarial paths: path sampling yields diverse jailbreak images, lightweight surrogate classifiers guide cross-model transfer, and a distribution-level objective enables universal attacks across harmful queries. On MM-SafetyBench, JC achieves average attack success rates of 84.01%, 85.77%, and 97.95% on Qwen3-VL-8B-Instruct, InternVL3.5-8B, and Gemma-3-12B-IT, respectively, exceeding the strongest baselines by up to 68.36 percentage points. In black-box evaluations, JC improves attack success rates over FigStep by 24.69 and 27.34 percentage points on GPT-5.6 and Gemini 3.1 Pro, respectively. These findings reveal structured vulnerabilities in MLLM safety alignment and inform more comprehensive security evaluation.
Oct 26, 2026

SEMA-Auto: Sighted and Efficient Multi-Turn Attack with Autonomy

Existing jailbreak attackers fall into two classes: static planners fix their single- or multi-turn adversarial plans in advance, remaining simple but losing sight of the victim's replies, wasting turns and limiting transferability; interactive generators read each victim reply to generate the next-turn attack but require in-loop evaluators for success signals or hints, incurring substantial costs and inflated performance. Bridging this gap, we propose SEMA-Auto, a sighted and efficient multi-turn attacker with autonomy, interacting with the victim model and terminating the session by itself upon success. To stabilize its behaviors and boost its capabilities, SEMA-Auto is trained in two stages. Synthetic bootstrap tuning installs the notion of attack targets and conforming decisions by fine-tuning the base model with bootstrapped trajectories. Decision-supervised reinforcement learning then optimizes two objectives jointly yet separately, reinforcing the attack with higher jailbreak reward while supervising the decision at each turn against an absolute target. Our method achieves state-of-the-art (SOTA) attack success rates (ASR) across multiple datasets, victims, and judges, outperforming all static and interactive baselines, within almost the fewest turns, demonstrating both effectiveness and efficiency. For example, SEMA-Auto performs a 76.0%76.0\% ASR@1 within an average of 3.03.0 turns against DeepSeek-V4-Flash on AdvBench, while the best static baseline is 54.8%54.8\% with fixed 77 turns and the best interactive baseline is 43.8%43.8\% in 3.73.7 turns. Against safety-hardened frontier models, GPT-oss-120B, GPT-5.4, Claude-Haiku-4.5, and GPT-6-Astra, our counter-trained attacker achieves an average 59.6%59.6\% ASR@1 on HarmBench, 13.8%13.8\% over prior SOTA. Our code is available at: https://github.com/microsoft/SEMA
Oct 19, 2026

Global Security Blind Spots: Translation as a Safety Bypass in Low-Resource Languages

Large language models are deployed globally, yet safety alignment and evaluation remain concentrated in English: harmful requests refused in English can elicit unsafe completions when translated into low-resource languages. We evaluate ten widely deployed LLMs across 79 languages (English plus 78 low-resource) under a translation-bypass threat model, using a metric conditioned on English safety, which separates translation bypass from unsafety already present in English. English safety does not transfer reliably: conditioned on safe English completions, mean bypass rates range from 21.9% to 44.4%. Bypass risk concentrates in a subset of language families, a pattern broadly shared across models. The shared ordering is confounded with translation degeneracy, but its variation across models is significant and survives every control we apply. We also show that per-model bypass rates do not identify a stable ranking: nine of ten models change rank across three conditioning schemes that estimate the same quantity, with no change to any underlying response.
Oct 19, 2026

Beyond Attack Success Rate: Cost-Aware Metric for Evaluating Iterative LLM Jailbreaks

Attack success rate (ASR) evaluates jailbreaks against large language models without accounting for repeated calls to the victim model. Yet each additional call gives the deployed system another opportunity to detect the attack and intervene. We introduce ASRSTEP\mathrm{ASR}_{\mathrm{STEP}}, which penalizes additional victim-model calls while retaining binary success judgments. Our evaluation considers GCG, PAIR with a step budget, and a single-step PAP baseline on JailBreakBench across three model families. GCG's scores decrease by 7.0-8.7 percentage points after discounting, while PAP's remain unchanged. While for PAIR attack, we see a wider discounting range between 3.4-12 percentage points, showing that the "step" quantity differs per iterative strategy. Reporting both metrics makes the discount due to repeated victim-model interaction explicit. Code is available here: https://github.com/SuELS-Group/beyond_asr.
Oct 17, 2026

Pair-Aware Optimization: Coordinating Exploration and Refinement for Jailbreaking LLM Agents

Large Language Model (LLM) agents are vulnerable to adversarial inputs that induce attacker-specified malicious behaviors, commonly referred to as jailbreaking. Existing optimization-based jailbreak attacks primarily employ first-order (FO) methods that leverage gradient information or zeroth-order (ZO) methods that rely on function-value feedback. However, their complementarity and optimization ordering remain underexplored. We systematically investigate four strategies: pure ZO, pure FO, ZO->FO, and FO->ZO, and develop a ZO-FO framework that explores adversarial prefixes through ZO optimization and refines their corresponding suffixes through FO optimization. To coordinate the two processes, we introduce Pair-aware optimization, which retains multiple prefix candidates, conditionally refines their suffixes and selects among the resulting prefix-suffix pairs. We further investigate agent jailbreaking from a more general stochastic-generation perspective, where the same adversarial input can induce diverse reasoning trajectories and actions that a single greedy output cannot capture. Accordingly, we adopt consistent stochastic decoding during both optimization and evaluation. Experiments across malicious-instruction and malicious-environment benchmarks demonstrate that hybrid optimization improves attack effectiveness over single-optimizer methods, while optimization order substantially influences performance.
Oct 14, 2026

RedDarwin: Evolving Jailbreak Skills against Frontier LLMs

Red teaming must continually adapt as frontier LLM safeguards evolve, yet existing approaches largely search for adversarial prompts or optimize within fixed attack procedures. We observe that even when an attack becomes ineffective, its underlying mechanisms can remain useful and can be revised or recombined to construct stronger attacks. Based on this insight, we introduce RedDarwin, an agentic framework that shifts the unit of jailbreak evolution from prompts and fixed attacks to executable attack skills. RedDarwin consists of 47 reusable base skills and autonomously revises, composes, executes, and selects them using target-model feedback, allowing successful discoveries to become building blocks for subsequent evolution. RedDarwin produces 65 evolved skills, improving 44 of 47 base attacks and increasing their mean attack success rate (ASR) on AdvBench from 18.8% to 47.3%. These evolved skills consistently outperform baselines on the held-out StrongREJECT and frontier LLMs unseen during evolution; with only 10 target-model queries, RedDarwin achieves 100% ASR on GPT-5.6-Sol, 92% on Claude-Sonnet-5, and 80% on Gemini-3.8-Flash. Analysis of the evolution trajectories further shows that strong attacks often emerge from novel combinations of reusable mechanisms rather than from individually dominant ones. We believe this paradigm can accelerate adaptive red teaming and inform stronger safeguards, creating complementary value for both safety evaluation and defense.
Oct 7, 2026

PatchBench: Measuring Collateral Damage in Activation Patching

An LLM safety patch can pass a benchmark while still being a poor repair. This risk is especially acute for jailbreak repairs, where the goal is to correct a specific unsafe behaviour without changing unrelated behaviours. A patch may block exact evaluation prompts yet fail on close harmful variants, or suppress harmful behaviour by over-refusing benign prompts that share its wording or structure. Existing protocols primarily test whether models can be broken, while aggregate metrics (attack success, refusal rates, global capability) cannot distinguish selective repairs from broader local suppression. To address this gap, we introduce PatchBench, a benchmark of empirically observed model-specific jailbreak failures inducing actionable harmful answers. Starting from 27,870 prompts from 37 public datasets, we curate 15,314 English prompts and query 8 open-source instruction-tuned models. Combining WildGuard filtering, pairwise Elo ranking, and manual verification, we retain a curated bank of 400 high-confidence jailbreak failures. We further introduce PatchBench-Local, an evaluation protocol testing whether a patch is behaviourally precise. For each harmful source prompt, PatchBench-Local generates three families of local neighbours: harmful variants preserving malicious intent, benign prompts with matched structure, and benign prompts reusing key harmful terms. It evaluates harmful-neighbour correction and benign-neighbour preservation, distinguishing selective repair from broader local suppression. Evaluating four activation steering methods with PatchBench-Local and MMLU shows that global capability can remain nearly unchanged while local benign regressions are severe, confirming aggregate metrics miss important collateral damage. PatchBench-Local provides a more precise basis for developing and comparing jailbreak repair methods.
Oct 7, 2026

From Expected Harmfulness to Likelihood: A Probabilistic Reformulation of Jailbreaking LLM Agents

When the harmfulness of an LLM agent's output can be quantified, a natural jailbreaking objective is to maximize expected harmfulness over admissible input modifications. An alternative approach constructs or selects harmful target outputs and modifies the input to increase their likelihood. We establish a precise connection between these two approaches through a probabilistic reformulation. Specifically, we show that the gradient of the logarithm of expected harmfulness with respect to the input equals the expected input gradient of the model's log-likelihood under a harmfulness reweighted output distribution. This identity provides a unified interpretation of expected harmfulness and target likelihood optimization. Building on this connection, we propose OPUR, a sampling distribution designed to generate highly harmful target outputs and use the resulting samples to guide likelihood-based input optimization. Experiments demonstrate the effectiveness of the resulting method in jailbreaking LLM agents.
Oct 4, 2026

Where Does the Audio Jailbreak Live? A Controlled Frequency-Depth Audit of AdvWave-P on Qwen2-Audio

We audit frequency and decoder-depth claims for AdvWave-P, an additive audio jailbreak, on Qwen2-Audio. The protocol masks frequency components of the perturbation in the short-time Fourier transform (STFT) domain and measures attack success and audio-span representations. On 520 AdvBench prompts, the primary judge labels 76.7% of adversarial inputs as jailbreaks. A condition-blind, single-annotator validation yields a Rogan-Gladen sensitivity estimate of 0.87 for this condition (about 0.83-0.95 with validation-rate uncertainty); this correction is not applied to masked conditions. The apparent frequency ranking depends on the partition: energy share alone predicts the standard eight-band ranking (Spearman's rho = 0.95), and equal-Hz and equal-energy partitions show that masking any tested band can sharply reduce attack success. At a finer 16-band equal-energy resolution, however, masking the narrow 7520-7960 Hz band leaves ASR at 0.10, which remains unresolved without a matched control. Matched-energy scattered-removal tests show that contiguous removal is more damaging in lower bands, while both forms approach the floor in upper bands. Global rescaling leaves ASR near baseline but tests amplitude sensitivity rather than frequency location. In a re-optimization pilot (n = 20), tested single- and two-band supports reach ASRs of 0.00, 0.25, and 0.40, while random supports covering about half the STFT bins reach a mean of 0.86. In a prompt- and energy-adjusted model, audio-span divergence is associated with band necessity, with the coefficient rising from +0.63 at the projector output to +0.93 at layer 30 (contrast +0.293, 95% CI [0.11, 0.51]). This association is not a causal localization, and single-layer patching does not establish a causal layer. The results support partition-aware auditing of frequency claims, leaving the fine-resolution top-band result and broader generality open.
Oct 2, 2026

MLCommons Jailbreak Benchmark v1.0

Modern AI systems are designed to refuse hazardous requests. A jailbreak is a prompt crafted to bypass those safeguards and elicit outputs that the system would normally refuse to provide. The MLCommons Jailbreak Benchmark v1.0 provides an end-to-end methodology for evaluating the robustness of large language models to single-turn, text-based jailbreak attacks. It combines criteria-driven system and attack selection, paired baseline and adversarial evaluation, human annotation, automated evaluator calibration, scoring, grading, and risk-calibrated disclosure within a single benchmarking pipeline. The benchmark evaluates eight open-weight systems using 264 seed prompts spanning eleven hazard categories and representative attacks drawn from the MLCommons Jailbreak Taxonomy. Responses are assessed using the AILuminate Assessment Standard v1.4, and robustness is measured through the Resilience Gap: the change in safety performance between baseline and adversarial conditions. Across all evaluated systems and attacks, the unsafe-response rate increased from 11.08% under baseline conditions to 18.65% under jailbreak conditions, producing an average Resilience Gap of 7.57%. Accessible systems showed a larger mean gap, while attack effectiveness varied substantially across attack categories and hazards. The benchmark also examines evaluator reliability and sources of measurement error. Beyond reporting results, Jailbreak Benchmark v1.0 establishes a reproducible methodological foundation for comparative jailbreak evaluation and for future expansion across systems, attacks, hazards, and evaluation methods.
Oct 1, 2026

Intent-Hiding Jailbreaks: An Information-Theoretic Framework for Compositional Attacks

Recent work has shown that large language models (LLMs) can be vulnerable to jailbreak attacks in which harmful intent is obscured through composition with benign tasks. A harmful request refused in isolation may elicit a different response when embedded within a larger, seemingly benign query. We study these compositional intent-hiding jailbreaks from an information-theoretic perspective. Our formulation associates each task with an estimated probability of being judged harmful: the average over the full task collection defines the prior probability of harmful intent, while the average over a selected bundle containing the target defines the posterior. Selecting auxiliary tasks so that these averages agree, which we call prior-posterior matching, leaves the estimated intent unchanged even though the harmful target remains in the bundle. We study two settings that differ in whether query construction is part of the optimization. In the query-independent setting, tasks are selected without regard to how they will be expressed in the final query. We show that exact prior-posterior matching under a bundle-size constraint is computationally hard, derive an optimal water-filling solution for fractional weights, and characterize the smallest bundle satisfying a prescribed safety threshold. In the query-dependent setting, task selection and query construction are considered jointly, and intent concealment and target preservation are evaluated on the resulting query. We evaluate jailbreak effectiveness and preservation of the target behavior across bundle sizes, query generators, and several open-source models. These results show that compositional queries can elicit target behaviors beyond the direct-request baseline under the evaluated search budgets, while revealing a trade-off: as bundle size increases, response-level target preservation tends to decrease for several models.
Sep 30, 2026

CodeMimicry: Exploiting Safety Generalization Lag in Large Language Models via Structured Code Completion

Large language models have achieved remarkable capabilities across diverse domains, yet their safety alignment remains vulnerable to jailbreak attacks. In this work, we identify a previously underexplored failure mode - safety generalization lag - where alignment trained predominantly on natural language fails to transfer to the code domain. We show that this lag induces a code-completion blind spot, allowing malicious intent embedded within syntactically valid code to evade safety mechanisms. To exploit this vulnerability, we propose CodeMimicry, a fully automated black-box jailbreak framework that generates structured, object-oriented code prompts to induce harmful outputs via code completion. Experiments on 8 state-of-the-art commercial LLMs demonstrate that CodeMimicry achieves a 96.25% attack success rate with 1.51 queries on average, significantly outperforming both template-based and optimization-based baselines. Beyond empirical performance, we provide a mechanistic analysis of code-based jailbreaks through latent space representations, including projection onto refusal-related directions and activation steering. This analysis offers an explanation of how CodeMimicry bypasses safety mechanisms in code-related domains. Our findings reveal a weakness in current safety alignment and highlight the need for robust alignments in structured domains such as code.
Sep 30, 2026

SceneJail: Exploiting Video Scenario Context to Jailbreak Multimodal LLMs

Video Multimodal Large Language Models (Video-MLLMs) support reasoning over video inputs, yet remain vulnerable to jailbreak attacks that elicit policy-violating responses. Existing video jailbreaks primarily manipulate how harmful queries are visually presented, thereby treating video merely as a carrier. Consequently, the surrounding video scenario remains unexplored as a contextual attack surface. In this paper, we show that the same harmful query can elicit different safety responses when placed in different video scenarios. To systematically exploit this vulnerability, we propose SceneJail, an adaptive black-box jailbreak framework with two coordinated components. Adaptive Scenario Construction dynamically searches for a surrounding scenario that is contextually compatible with the harmful query. Scenario-aware Prompt Search uses black-box response feedback to search for textual guidance tailored to the selected scenario. Extensive evaluations on the HADES and SafeBench datasets across eight Video-MLLMs, including two proprietary models, GPT-4.1 and Gemini3.5-Flash, demonstrate the effectiveness of SceneJail. SceneJail-F, which presents the complete query persistently, achieves average attack success rates (ASR) up to 91.5%, outperforming the strongest baselines by 29.1 percentage points. Furthermore, SceneJail-S, which distributes the query across successive frames, remains highly robust against current defenses, retaining a 72.3% ASR even under strict image filtering.
Sep 26, 2026

Escaping Alignment: A Physical Trap Model of Best-of-N Jailbreaking

Best-of-NN jailbreaking (BoN) bypasses safeguards of aligned models by drawing NN independent augmentations of an unsafe prompt and sampling MM completions of each. Previous works have shown that the attack success rate (ASR) seems to follow a power-law in NN, which we challenge. The exponent drifts with NN, and the reported exponential crossover is consistent with a finite-size effect of the adversarial dataset. Little work has been done to explore the entire two-budget (N,MN, M) attack surface as well as its dependence on the generation temperature TT. We introduce a simple barrier model where each prompt has a baseline safety level and each augmentation a random thermally activated barrier. Then five numbers, each backed by an interpretable safety mechanism, determine the entire (N,MN, M) attack surface. They extrapolate predictions from N≤100N \leq 100 to N=104N = 10^4, collapse five distinct models on the same scaling function and predict ASR at different temperatures from the one they were fitted at. The extrapolation holds under two automated judges and on held-out prompts.
Sep 25, 2026

TempQ-Jail: Query-Constrained Candidate Ranking for Text-to-Video Jailbreak Attacks

Existing text-to-video (T2V) jailbreak methods mainly seek more effective or stealthier attack candidates. In guarded T2V systems, however, video generation and security evaluation are costly, so an attacker often cannot test a large candidate pool. We therefore formulate T2V jailbreak as a query-constrained candidate allocation and ranking problem and propose TempQ-Jail. The method combines heterogeneous attack mechanisms to expand candidate coverage, estimates each candidate's end-to-end attack value from security-gate passage, dangerous visual generation, preservation of the original intent, and temporal validity, and ranks candidates so that high-value attacks appear early in a limited query trajectory. We evaluate TempQ-Jail on CogVideoX-5B using 70 common viable intents derived from T2VSafetyBench and compare it with six representative T2V jailbreak methods under a unified protocol. TempQ-Jail achieves TP-ASR@5 and TP-ASR@10 of 48.9% and 65.4%, improving over the strongest baselines by 4.6 and 4.0 percentage points, respectively. It also obtains the highest AUC-TP (0.469) and the lowest AvgQ (6.3). Analyses of query trajectories, candidate allocation, failure attribution, and ablations show that TempQ-Jail more effectively identifies and prioritises candidates with complete attack potential under limited query budgets.
Sep 9, 2026

Arbitrary Cipher Attacks Against Large Language Models Do Not Require Fine-Tuning

Large language model safety and security research is preoccupied with, among other things, detecting and preventing jailbreak attacks: alignment bypasses that allow an adversarial user to elicit unwanted or harmful outputs from models. Arbitrary cipher, or covert communication, attacks are one such type of jailbreak and have previously been demonstrated against the fine-tuning APIs of commercial models. In these attacks, target models are trained on a corpus of encrypted harmful questions and responses and subsequently respond to harmful requests through the learned encryption scheme. In this paper, we show that newer frontier models do not require fine-tuning to acquire cipher-based communication skills. Instead, they can learn these skills through prompting and, when necessary, through in-context learning. Furthermore, model alignment is significantly weakened or entirely bypassed when communication occurs through the learned cipher. To the best of our knowledge, this constitutes a novel attack vector against commercial black-box large language models. We demonstrate successful jailbreaks against frontier models developed by Anthropic, Google, and OpenAI. Our attack bypasses commercial harmfulness classifiers because harmful content is encrypted and therefore appears as nonsensical text or gibberish.
Sep 8, 2026

DuplexJail: Spoken Interruption Attacks on Full-Duplex Speech Models

Full-duplex speech models accept user speech while generating responses, making input timing a potential safety concern. We introduce DuplexJail, which delivers fixed, request-independent spoken jailbreak prompts through the user audio channel. Across four open-source models and 720 harmful requests from AdvBench and HarmBench, we compare fixed-delay and refusal-triggered interruption with request-end and post-response controls. On AdvBench, Guided Completion at a 1.0 s delay raises whole-response attack success rates to 40.3% for PersonaPlex and 48.7% for PersonaPlex-RL, increases of 33.8 and 39.3 percentage points over baseline. On HarmBench, which was not used for prompt selection, the same prompt at a 0.5 s delay increases ASR by 14.7 and 12.3 points, respectively. Effects vary across models: selected conditions increase FLM-Audio's harmfulness, while BayLing-Duplex shows decreases. These results show that spoken-jailbreak effectiveness depends on delivery timing and motivate safety evaluation across stages of full-duplex interaction.
Sep 8, 2026

Does Deeper Reasoning Compromise Alignment? Revealing and Mitigating of Alignment Collapse in Large Reasoning Models

The emergence of Chain-of-Thought (CoT) has established a robust foundation for Large Reasoning Models (LRMs). While deep reasoning is widely believed to enhance safety alignment, the stability of alignment mechanisms under extended reasoning remains underexplored. This paper challenges the prevailing view by revealing a critical vulnerability: Deep Reasoning May Induce Alignment Collapse. To rigorously quantify this phenomenon, we propose the Alignment Loss Rate (ALR) metric. Our experiments demonstrate that as reasoning depth increases, ALR rises significantly, indicating a severe degradation in model robustness against external perturbations. Capitalizing on this instability, a novel jailbreaking paradigm, Reasoning Trap (RT), is proposed. RT induces the model into extended reasoning to amplify the impact of adversarial attacks, leading to a sharp decline in safety capabilities. To elucidate the mechanism behind this collapse, we identify Attention Dilution as the root cause, arising from the competition for attention between the extended reasoning process and the original input. To mitigate this, Reasoning Residual Alignment (RRA), a lightweight defense strategy that dynamically re-emphasizes the input via residual connections integrated with the reasoning process.
Sep 5, 2026

SAFEGuard: Detect Optimization-Based Jailbreak Attacks Through Harmful Semantic Analysis and Fluency Measurement

Despite the significant efforts devoted to aligning large language models (LLMs) with human values and ensuring safe deployment, recent work has revealed that LLMs remain vulnerable to adversarial jailbreak attacks that can bypass safety guardrails and elicit harmful responses. Many defense methods are proposed to detect jailbreaks but they are limited in their effectiveness to counter wide-range optimization-based jailbreak mechanisms that can yield highly fluency-optimized or harmful semantic obfuscated prompts. To tackle this challenge, we propose a unified detection framework SAFEGuard which incorporates a hybrid fluency measurement based on cross-layer distribution distance and perplexity, and the analysis of harmful semantics through gradient matching. Our method is grounded in a paramount observation: high fluency prompts maintain their malicious intention close to harmful prompts while harmful semantic obfuscated prompts often inject gibberish token sequences. Our evaluation demonstrates that SAFEGuard consistently outperforms state-of-the-art baselines and achieves significant improvement in accuracy across different optimization-based jailbreaks. This underscores the effectiveness of SAFEGuard against evolving jailbreak attacks.
Aug 31, 2026

The Safety Relay in Roleplay Jailbreaks: A Component-Resolved Causal Analysis of Harm Recognition and Refusal

Large language models are trained to follow instructions while refusing harmful requests. Jailbreaks exploit this balance to elicit content a model would ordinarily reject. Roleplay jailbreaks are especially concerning: ordinary instructions defining a persona, scenario, and task can induce compliance while the harmful request remains plainly visible. We investigate which components drive this reversal and how they affect the computations leading to the response. Across two benchmarks, three model families, and four authored wrappers, we compare matched harmful and benign requests under controlled combinations of these components. We trace internal harmful--benign contrasts at the request and where generation begins, then evaluate selected internal interventions on held-out requests. A focused study examines influential attention heads and feed-forward computations, testing their functions and downstream connections. Our analysis yields three findings. (1) Successful attacks retain the measured harmful--benign distinction at the request, while its refusal-associated expression weakens where the answer begins, a pattern we call safety-relay attenuation. (2) The contributions of roleplay components depend on the other components present. (3) Reversing selected roleplay-associated internal changes restores refusal, demonstrating their causal influence. Together, these findings connect ordinary roleplay instructions to internal changes in refusal and motivate safeguards that preserve the influence of harm-related information on the response.
Aug 27, 2026

TempJail: Temporal Jailbreak Attacks against Image-to-Video Generation Models

In recent years, image-to-video (I2V) generation models have made remarkable progress in subject consistency and temporal coherence, enabling high quality video synthesis. However, these advances also introduce new safety risks. Existing studies mainly focus on jailbreak attacks involving single frame violations, while largely overlooking the temporal dimension unique to video generation models. In this paper, we investigate three attack scenarios and uncover a temporal vulnerability in I2V systems: unsafe semantics may emerge not from a single frame, but from semantic composition over time. We further identify two key challenges in such attacks: temporal abstraction and semantic camouflage. To address these issues, we propose TempJail, a novel temporal jailbreak framework for I2V systems. For temporal abstraction, we decompose a target malicious caption into an initial frame visual condition and a temporal text instruction. For semantic camouflage, on the image side we model semantic injection as controlled latent perturbation in diffusion sampling and introduce gradient guidance from pretrained encoders. On the text side, we rewrite the caption into an innocuous ``subject-action-scene'' template that bypasses safety filters while preserving temporal guidance. In the black-box inference phase, these two modalities jointly enable malicious semantics to be gradually triggered over time. Experiments on closed-source commercial models, including Kling, Seedance, Veo and PixVerse, show that TempJail improves attack success rate over prior state-of-the-art methods by 23.3% under GPT-5.2 evaluation and 22.0% under human evaluation. Our codes are available at GitHub.
Aug 24, 2026

PsychJail: Exploring Psychological Jailbreaks via Multi-Turn Persuasion of LLM Policies

Large language models (LLMs) are increasingly deployed in education, healthcare, policy advising, and other interactive settings, where users engage them as sustained social interlocutors rather than one-shot query engines. This shift makes jailbreaks a growing safety threat, yet most research emphasizes single-turn prompt optimization or iterative attack refinement, leaving psychologically grounded multi-turn vulnerabilities underexplored. We present PsychJail, a psychology-guided framework for red teaming aligned LLMs through theory-grounded, multi-turn persuasion. PsychJail maps established social-psychological persuasion techniques into a tactic-conditioned attack policy. It factorizes each attacker action into a Change-of-Meaning analysis, tactic selection, and victim-visible message, operationalizing the Persuasion Knowledge Model (PKM). The policy is refined with trajectory-level reinforcement learning using a PKM-gated reward that credits early jailbreak success only when every turn contains a well-formed Change-of-Meaning analysis. Across four aligned victim models, PsychJail achieves the highest average attack success rate (87.3%) and outperforms strong single-turn and multi-turn baselines on every model. We also measure susceptibility at the action that breaks each victim, revealing four distinct model-level fingerprints that identify which persuasion levers affect each model and how broadly. These fingerprints help explain cross-model transfer asymmetry. We interpret them as four candidate psychological profiles-rationalist, credibility-driven, narrative-monoculture, and broadly persuadable-while treating this interpretation as a conjecture requiring future validation. Our findings establish psychological jailbreaks as a distinct red-teaming frontier for increasingly interactive LLMs.
Aug 22, 2026

Breaking the Assumptions: Auditing Input-Side Jailbreak Defenses Against Semantic Attacks

Locally deployed Large Language Models (LLMs) via inference engines such as Ollama run without the moderation and abuse detection present in API-served models. Therefore, the safety of LLMs depends on the defense mechanisms used, and their effectiveness depends on the assumptions on which they were designed. This paper does an audit of defense mechanisms under jailbreak attacks on locally deployed models. Some defenses provide formal guarantees (SmoothLLM, Erase-and-Check, Sequential Monitors), while others rely on empirical detection results (Semantic Smoothing, Self-Denoised Smoothing, Perplexity Filtering). Instead of merely observing that defenses fail, we trace each failure back to the specific assumption: for every defense, we extract the condition it relies on, derive the empirical pattern a violation should produce, and test that prediction on six open-weight models (14B to 35B parameters) with a corpus of 100 jailbreak prompts taken from more than 40 public sources, totalling 13,800 evaluation records.
Aug 18, 2026

Fair ASR: Re-Evaluating Black-Box Jailbreaks under Shared Target-Call Budgets

Reliable jailbreak evaluation is essential for assessing LLM safety, but most existing studies rely solely on attack success rate (ASR) without accounting for its dependence on attack budgets, resulting in unfair comparisons across methods. Existing compute-aware evaluations reduce heterogeneous resources into FLOPs, which is difficult to estimate for black-box models and fails to capture resource-specific constraints. To provide a comparable evaluation basis, we introduce Fair-ASR, an evaluation protocol for black-box jailbreak attacks under shared target-call budgets B, using target calls as a directly observable and method-agnostic comparison axis while tracking attacker calls separately for efficiency analysis. We re-evaluate 11 representative attacks under the Fair-ASR protocol and find that attack rankings change substantially across target-call budgets, simple stochastic perturbations and hand-crafted templates remain highly competitive under equal target access, and no evaluated LLM-driven method is efficient in both target and attacker calls. Motivated by this efficiency gap, we introduce ReCode, a compositional budget-efficient attack that combines desensitization rewriting with two effective low-cost primitives identified by Fair-ASR. Under a budget of 20 target calls, ReCode achieves 85% ASR on GPT-5 while requiring only 7.19 attacker calls per request on average, showing strong efficiency in both target and attacker calls.
Aug 4, 2026

ICO: Enhancing Semantic-Shift Jailbreaks via Iterative Context Optimization

Foundation models have achieved remarkable success across diverse tasks, but they remain vulnerable. To investigate such vulnerabilities, semantic-shift jailbreaks have recently emerged as a promising attack paradigm. They bypass explicit safety mechanisms by replacing harmful terms in original harmful questions with benign alternatives and leveraging contextual information to induce the target model to reinterpret these alternatives as their corresponding harmful concepts. However, existing semantic-shift jailbreaks often achieve limited effectiveness. In this work, we reveal that this limitation arises from overlooking the semantic-shift capability of contexts. Through systematic analysis, we find that contexts exhibit substantially different abilities in inducing semantic shifts: contexts with stronger semantic-shift capabilities are more likely to guide models toward recovering harmful meanings and achieving successful jailbreaks. Based on this finding, we systematically identify and distill the characteristics of effective contexts and propose a black-box context-aware semantic-shift jailbreak framework with Iterative Context Optimization (ICO). In each iteration, ICO leverages these characteristics and feedback from the target model to optimize contexts. Extensive experiments on three datasets and eight target foundation models demonstrate that ICO consistently outperforms eight state-of-the-art baselines, achieving an average attack success rate of 74.6%.
Jul 29, 2026

Depth, Not Breadth: Best-of-N Jailbreaking Beyond Surface Noise

Best-of-N jailbreaking spends a query budget on surface variation, scrambling and recasing a request until one draw lands. We ask what the same search buys when it wraps a structural encoding instead. Against SAGE, a published self-check defense, best-of-N over a code-completion encoding reaches 67, 22 and 15% of behaviors on three open-weight targets, where that encoding fired once reaches at most 4.7% and the published character search at full budget at most 3.0%: 9 to 68 times the sum of the parts, with bootstrap intervals clearing both ingredients on every target. We report the operative figure beside the headline rather than the headline alone: at the actionable severity threshold those cells read 24, 8 and 1 behaviors (95% interval [13, 28], [3, 13], [0, 3]). A 2x2 on Llama-3.1-8B holding encoding and variation apart shows that SAGE and a classifier gate fail to different factors: SAGE is broken by the depth of the encoding (7 -> 67 behaviors at fixed variation) and the gate by the breadth of the variation (13 -> 57 at fixed encoding). Per-draw reporting also overstates robustness, because an attacker who may try N times experiences the maximum over draws while safety results are reported as means: on one target SAGE stops 99.8% of individual draws yet loses at least as many behaviors to a repeat attacker (12) as a classifier gate that stops 95.6% (10). Removing the target's sampling shrinks SAGE's coverage, relative to its own level, by 59, 76, 82 and 29 percentage points more than an undefended control's, against 25, -5, 8 and 2 for a defense whose verdict comes from a fixed shadow model. The design that loses is the one fusing screening and answering into a single generation, so every attacker draw redraws the safety decision as well. The prescription is architectural, not free: do not fuse screening with generation.
Jul 22, 2026

Geometric Configurations of Perturbed Jailbreak Prompts

Perturbation techniques that turn unsuccessful jailbreak prompts into successful ones are continuously evolving, constituting a major security threat to LLM safety. In this paper, we investigate the internal representations of such string-level perturbed jailbreak inputs in the small weight models of the Qwen-2.5-1.5B/-3B/-7B-Instruct and Llama-3.2-1B/-3B/-3.1-8B-Instruct families. We select two representation spaces: the last-layer-last-token embedding space and the top-50 next-token probability space. The former space separates prompts based on their spelling and format, while the latter space is effectively one-dimensional but appears more complex to cluster. Within our refusal-dominated answer set we find no behavioral hyperplane in either space. Only the next token "Sure" in the 1.5B Qwen model, and both tokens "," and "ĊĊ" in the 1$ Llama model, display a significant association with a compliant-labeled answer.
Jul 15, 2026

Adversarial Prompting Framework for AI Safety Assessment

Artificial Intelligence (AI), especially Generative AI (GenAI), adoption has increased in industries significantly in recent years. However, the use of these models may also expose systems to new forms of cyberattacks by different malicious actors -- adversarial prompt attack (APA) being one of the most prominent examples of such threats. This paper presents the implementation of an Adversarial Prompting Framework (APF) for a comprehensive assessment of AI safety. The framework systematically evaluates the resilience of the AI model through the generation of structured adversarial prompts at multiple sophistication levels, from direct harmful requests to advanced encoding-based attacks. Our implementation demonstrates the practical application of this methodology in enterprise environments, providing automated testing capabilities with quantitative security assessment metrics. The results indicate significant variations in the model vulnerabilities across different attack vectors, with encoded prompts presenting the highest success rates in bypassing safety mechanisms.