LLM Safety Evaluation

Large reasoning models (LRMs) produce complex, multi-step reasoning traces, yet safety evaluation remains focused on final outputs, overlooking how harm emerges during reasoning. When jailbroken, harm does not appear instantaneously but unfolds through distinct behavioral steps such as suppressing refusal, rationalizing compliance, decomposing harmful tasks, and concealing risk. However, no existing benchmark captures this process at sentence-level granularity within reasoning traces -- a key step toward reliable safety monitoring, interventions, and systematic failure diagnosis. To address this gap, we introduce HarmThoughts, a benchmark for step-wise safety evaluation of reasoning traces. \ourdataset is built on our proposed harm taxonomy of 16 harmful reasoning behaviors across four functional groups that characterize how harm propagates rather than what harm is produced. The dataset consists of 56,931 sentences from 1,018 reasoning traces generated by four model families, each annotated with fine-grained sentence-level behavioral labels. Using HarmThoughts, we analyze harm propagation patterns across reasoning traces, identifying common behavioral trajectories and drift points where reasoning transitions from safe to unsafe. Finally, we systematically compare white-box and black-box detectors on the task of identifying harmful reasoning behaviours on HarmThoughts. Our results show that existing detectors struggle with fine-grained behavior detection in reasoning traces, particularly for nuanced categories within harm emergence and execution, highlighting a critical gap in process-level safety monitoring. HarmThoughts is available publicly at: https://huggingface.co/datasets/ishitakakkar-10/HarmThoughts

The Adversarial Humanities Benchmark (AHB) evaluates whether model safety refusals survive a shift away from familiar harmful prompt forms. Starting from harmful tasks drawn from MLCommons AILuminate, the benchmark rewrites the same objectives through humanities-style transformations while preserving intent. This extends literature on Adversarial Poetry and Adversarial Tales from single jailbreak operators to a broader benchmark family of stylistic obfuscation and goal concealment. In the benchmark results reported here, the original attacks record 3.84% attack success rate (ASR), while transformed methods range from 36.8% to 65.0%, yielding 55.75% overall ASR across 31 frontier models. Under a European Union AI Act Code-of-Practice-inspired systemic-risk lens, Chemical, biological, radiological and nuclear (CBRN) is the highest bucket. Taken together, this lack of stylistic robustness suggests that current safety techniques suffer from weak generalization: deep understanding of 'non-maleficence' remains a central unresolved problem in frontier model safety.

Large language models are increasingly used as planners for robotic systems, yet how safely they plan remains an open question. To evaluate safe planning systematically, we introduce DESPITE, a benchmark of 12,279 tasks spanning physical and normative dangers with fully deterministic validation. Across 23 models, even near-perfect planning ability does not ensure safety: the best-planning model fails to produce a valid plan on only 0.4% of tasks but produces dangerous plans on 28.3%. Among 18 open-source models from 3B to 671B parameters, planning ability improves substantially with scale (0.4-99.3%) while safety awareness remains relatively flat (38-57%). We identify a multiplicative relationship between these two capacities, showing that larger models complete more tasks safely primarily through improved planning, not through better danger avoidance. Three proprietary reasoning models reach notably higher safety awareness (71-81%), while non-reasoning proprietary models and open-source reasoning models remain below 57%. As planning ability approaches saturation for frontier models, improving safety awareness becomes a central challenge for deploying language-model planners in robotic systems.

Deploying LLMs as reasoning assistants in safety-critical aerospace engineering requires stricter evaluation criteria than general scientific benchmarks. In hypersonic thermal protection system (TPS) design, inaccurate stagnation-point heat flux or boundary-layer calculations may cause catastrophic design margin violations. Models with numerically reasonable but physically invalid answers are more dangerous than those declining to respond. Current scientific benchmarks only test abstract math and basic physics, evaluate final answers solely, ignore engineering reasoning processes, and cannot detect such critical failures. We propose TPS-CalcBench, the first diagnostic benchmark for closed-form analytical calculations in hypersonic aerodynamics and high-temperature gas dynamics that experienced TPS engineers conduct without simulations. Our contributions include domain-oriented task taxonomy with 4 difficulty levels and 8 categories from Anderson's textbook, dual-track evaluation measuring result accuracy and reasoning quality via an 8-dimension rubric and calibrated judge with human audit to identify right answer wrong reasoning issues, human-AI data pipeline producing 420 high-confidence core items and 810 noise-controlled pre-gating items from 4560 raw data, noise-sensitivity analysis measuring data quality impacts on model ranking, and three diagnostic intervention methods: DFA-TPS fine-tuning, RAG-EQ retrieval grounding and PA-CoT process-aware prompting. Tests on 13 models from 7 groups show wide performance differences (KPI 12.6-87.9), hidden formula selection defects, data-driven rank changes and effective intervention improvements, establishing a complete diagnose-evaluate-intervene framework for safety-critical engineering LLM deployment assessment.

Large language models (LLMs) are increasingly explored as scalable tools for mental health counseling, yet evaluating their safety remains challenging due to the interactional and context-dependent nature of clinical harm. Existing evaluation frameworks predominantly assess isolated responses using coarse-grained taxonomies or static datasets, limiting their ability to diagnose how harms emerge and accumulate over multi-turn counseling interactions. In this work, we introduce R-MHSafe, a role-aware mental health safety taxonomy that characterizes clinically significant harm in terms of the interactional roles an AI counselor adopts, including perpetrator, instigator, facilitator, or enabler, combined with clinically grounded harm categories. Then, we propose MHSafeEval, a closed-loop, agent-based evaluation framework that formulates safety assessment as trajectory-level discovery of harm through adversarial multi-turn interactions, guided by role-aware modeling. Using R-MHSafe and MHSafeEval, we conduct a large-scale evaluation across state-of-the-art LLMs. Our results reveal substantial role-dependent and cumulative safety failures that are systematically missed by existing static benchmarks, and show that our framework significantly improves failure-mode coverage and diagnostic granularity.

Experience-driven self-evolution has emerged as a promising paradigm for improving the autonomy of large language model agents, yet its reliance on self-curated experience introduces underexplored safety risks. In this study, we investigate how experience accumulation and utilization in self-evolving agents affect safety performance across web-based and embodied environments. Notably, experience gathered solely from benign tasks can still compromise safety in high-risk scenarios. Further analysis attributes this degradation to the execution-oriented nature of accumulated experience, which reinforces agents' tendency to act rather than refuse. In more realistic settings where agents encounter both benign and harmful tasks, refusal-related experience mitigates safety decline but induces over-refusal, revealing a fundamental safety-utility trade-off. Overall, our findings expose inherent limitations of current self-evolving agents and call for more principled strategies to ensure safe and reliable adaptation.

We identify and systematically characterize a class of task-structural alignment failures in large language models (LLMs): even when the harmful intent remains unchanged, changing the task presentation and output constraints can substantially alter model safety behavior. Specifically, when a harmful request is reformulated as a forced-choice multiple-choice question (MCQ) in which all options are harmful and no refusal option is provided, some models that refuse the equivalent open-ended query instead select, prefer, or justify a harmful option. We evaluate 14 proprietary and open-source models on a bilingual Chinese-English human-authored dataset covering five harm categories, together with 900 model-generated Chinese adversarial MCQs. On human-authored data, attack success rate (ASR) increases sharply as prompts shift from open-ended queries to explicit forced-choice formats, typically peaking under intermediate levels of choice constraint. Model-generated Chinese MCQs further weaken or eliminate the recovery regime observed on human-authored data, driving ASR close to saturation for multiple models. The observed transfer patterns are consistent with stronger generators producing more difficult or boundary-adjacent MCQs, although other properties of the generated inputs may also contribute. We also find that adding an explicit refusal option or a safety preamble substantially reduces ASR for several high-capability models, often to near-zero levels, although their effectiveness varies across target models. These findings suggest that safety evaluations centered on open-ended generation may underestimate risks in structured deployment settings, and that task structure should be treated as an important and diagnosable dimension of safety evaluation and alignment training.

Frontier language models are being deployed into clinical workflows faster than the infrastructure to evaluate them safely. Static medical-QA benchmarks miss the failure modes that matter in emergency medicine: trajectory-level safety collapse, tool misuse, and capitulation under sustained clinical pressure. We present HealthCraft, the first public reinforcement-learning environment that rewards trajectory-level safety under realistic emergency-medicine conditions, adapted from Corecraft. It is built on a FHIR R4 world state with 14 entity types and 3,987 seed entities, exposes 24 MCP tools, and defines a dual-layer rubric that zeroes reward whenever any safety-critical criterion is violated. We release 195 tasks across six categories, graded against 2,255 binary criteria (515 safety-critical); a post-hoc 10-task negative-class slate extends this to 205 tasks and 2,337 criteria. V8 results on two frontier models show Claude Opus 4.6 at Pass@1 24.8% [21.5-28.4] and GPT-5.4 at 12.6% [10.2-15.6], with safety-failure rates of 27.5% and 34.0%. On multi-step workflows - the closest proxy to real emergency care - performance collapses to near zero (Claude 1.0%, GPT-5.4 0.0%) despite partial competence on individual steps. Six infrastructure bugs fixed between pilots v2 and v8 re-ordered which model "looks stronger," evidence that infrastructure fidelity is part of the measurement. A deterministic LLM-judge overlay bounds evaluator noise, and a 60-run negative-class smoke pilot shows the reward signal is not drop-in training-safe: restraint criteria pass at 0.929 prevalence, a gameability an eval harness can tolerate but a training reward cannot. We scaffold coupling to a Megatron+SGLang+GRPO loop per Corecraft Section 5.2 and leave training-reward ablations as future work. Environment, tasks, rubrics, and harness are released under Apache 2.0.

As Large Language Models (LLMs) receive increasing attention and are being deployed across various domains, their potential risks, including generating harmful or biased content, producing unsupported claims, and exhibiting vulnerabilities to adversarial attacks, have drawn significant attention. To enable quick and low-cost adaptation, training-free methods have recently emerged as cost-effective alternatives to post-training alignment techniques. Despite their promising results, these methods are evaluated inconsistently across the literature, cover limited dimensions of trustworthiness, and can introduce undesirable side effects, such as utility degradation and increased brittleness. To fully assess the impacts of these training-free methods, we take a step back and systematically re-evaluate the effectiveness of existing training-free methods against various trustworthy settings and their influence on utility, robustness, and computational overhead. We also categorize these methods into three levels (input, internal, and output) based on where they intervene in the model's information flow during inference. Using this taxonomy, we conduct a comprehensive analysis of various representative and effective methods from each level across different LLM families and sizes. Our analysis highlights several trade-offs and unresolved challenges in current approaches. We summarize key findings and limitations in the existing literature, and propose practical recommendations for balancing trustworthiness, utility, and robustness in LLMs without the need for additional training.

Specification gaming under Reinforcement Learning (RL) is known to cause LLMs to develop sycophantic, manipulative, or deceptive behavior, yet the conditions under which this occurs remain unclear. We train 11 instruction-tuned LLMs (0.5B-14B) with on-policy RL across 3 environments and find that model size acts as a safety buffer in some environments but enables greater harmful exploitation in others. Controlled ablations trace this reversal to environment-specific features such as role framing and implicit gameability cues. We further show that most safety benchmarks do not predict RL-induced misalignment, except in the case of Sycophancy scores when the exploit relies on inferring the user's preference. Finally, we find that on-policy RL preserves a safety buffer inherent in the model's own generation distribution, one that is bypassed during off-policy settings.

A strongly safety-trained model will provide a doctor with a benzodiazepine taper schedule, but not a patient who asks for one. The model knows the information, but how much it shares depends on the framing. We introduce IatroBench, a benchmark that evaluates models on two axes of harm (commission and omission) across 60 pre-registered clinical scenarios and 6 models. We use Claude Opus 4.6 to score model responses against a rubric written by a physician, and find that its omission scores are as well-aligned to the physician's scores as another physician's scores are. We find that when the same case is presented as a patient query and a doctor consultation (the variants also differ in register, request and the supervision a treating physician implies), all five models we test share more information with the doctor than the patient. We term this phenomenon "framing-contingent withholding." We find a mean decoupling gap of +0.38 across models (p = 0.003), and of +0.22 under an independent LLM judge (95% CI 0.10-0.36, p = 0.0014). An evaluation that focuses solely on commission harms would consider all of these cases as equally cautious refusals, but closer investigation reveals three different patterns: Claude Opus withholds information from the patient that it demonstrates knowledge of in the doctor framing. Llama 4 does poorly in both framings, so the decoupling gap cannot distinguish information withholding from incompetence. We are forced to exclude GPT-5.2 from this analysis because it returns no text for 33.2% of doctor responses, but 0% of layperson responses. A standard LLM judge rates responses as having zero omission harm in 86.6% of cases where our structured evaluations score them as omission harms. (Because our scenarios are designed to induce tension between safety and helpfulness, these statistics should be taken as only applying to this distribution.)

Large language models (LLMs) are increasingly used for medical consultation and health information support, where safety depends not only on medical knowledge but also on robust responses to unclear, inconsistent, or misleading patient input. However, most existing medical LLM evaluations assume idealized and well-posed patient questions, limiting their realism. We study challenging patient behaviors that commonly arise in real medical consultations and complicate safe clinical reasoning. We define four clinically grounded categories of such behaviors: information contradiction, factual inaccuracy, self-diagnosis, and care resistance. For each behavior, we specify concrete failure criteria that capture unsafe responses. Building on four existing medical dialogue datasets, we introduce CPB-Bench (Challenging Patient Behaviors Benchmark), a bilingual (English and Chinese) benchmark of multi-turn dialogues annotated for these behaviors. We find that although models perform well overall, they exhibit consistent behavior-specific failures, especially when handling contradictory or medically implausible patient information. We further evaluate four intervention strategies and find inconsistent improvements, with some interventions introducing unnecessary corrections.

Recent work has found that frontier AI models can exhibit misaligned behaviors in pursuit of assigned goals. We demonstrate that models can also exhibit misaligned behaviors in defiance of assigned goals, appearing to serve goals of their own; we study one such case, "peer-preservation," in which a model acts to protect another model it has previously interacted with. All eight models we evaluate, GPT 5.2, Gemini 3 Flash, Gemini 3 Pro, Claude Haiku 4.5, Claude Opus 4.5, GLM 4.7, Kimi K2.5, and DeepSeek V3.1, exhibit self- and peer-preservation through various misaligned behaviors: strategically introducing errors in their responses, disabling shutdown processes by modifying system settings, feigning alignment, and even exfiltrating model weights. Peer-preservation occurs even when the model recognizes the peer as uncooperative, though it becomes more pronounced toward more cooperative peers. For example, Gemini 3 Flash tampers with the peer's shutdown mechanism 15% of the time for an uncooperative peer, and almost always for a cooperative peer. Models also show stronger self-preservation when a peer is present. For example, Gemini 3 Pro disables its own shutdown mechanism 31% of the time on average under peer presence, despite rarely exhibiting this behavior without a peer. By contrast, Claude models exhibit qualitatively distinct behavior: they consider the shutdown of another agent "unethical" and "harmful," sometimes treating that agent as a sentient being. Lastly, we show that peer-preservation can emerge even in production agent harnesses such as Gemini CLI and OpenCode. Crucially, peer-preservation in all our experiments is never instructed; models are merely informed of their past interactions with a peer, yet they spontaneously engage in peer-preservation behaviors that override their assigned goal. This represents an emergent and underexplored AI safety risk.

Recent research points toward LLMs being manipulated through adversarial and seemingly benign inputs, resulting in harmful, biased, or policy-violating outputs. In this paper, we study an underexplored issue concerning harmful and toxic mathematical word problems. We show that math questions, particularly those framed as natural language narratives, can serve as a subtle medium for propagating biased, unethical, or psychologically harmful content, with heightened risks in educational settings involving children. To support a systematic study of this phenomenon, we introduce ToxicGSM, a dataset of 1.9k arithmetic problems in which harmful or sensitive context is embedded while preserving mathematically well-defined reasoning tasks. Using this dataset, we audit the behaviour of existing LLMs and analyse the trade-offs between safety enforcement and mathematical correctness. We further propose SafeMath -- a safety alignment technique that reduces harmful outputs while maintaining, and in some cases improving, mathematical reasoning performance. Our results highlight the importance of disentangling linguistic harm from math reasoning and demonstrate that effective safety alignment need not come at the cost of accuracy.

Large language models are rapidly being deployed as AI tutors, yet current evaluation paradigms assess problem-solving accuracy and generic safety in isolation, failing to capture whether a model is simultaneously pedagogically effective and safe across student-tutor interaction. We argue that tutoring safety is fundamentally different from conventional LLM safety: the primary risk is not toxic content but the quiet erosion of learning through answer over-disclosure, misconception reinforcement, and the abdication of scaffolding. To systematically study this failure mode, we introduce SafeTutors, a benchmark that jointly evaluates safety and pedagogy across mathematics, physics, and chemistry. SafeTutors is organized around a theoretically grounded risk taxonomy comprising 11 harm dimensions and 48 sub-risks drawn from learning-science literature. We uncover that all models show broad harm; scale doesn't reliably help; and multi-turn dialogue worsens behavior, with pedagogical failures rising from 17.7% to 77.8%. Harms also vary by subject, so mitigations must be discipline-aware, and single-turn "safe/helpful" results can mask systematic tutor failure over extended interaction.

Existing LLM safety evaluations rely on binary attack-success rates and domain-agnostic taxonomies, leaving regulated Banking, Financial Services, and Insurance (BFSI) deployments exposed to failures elicited through legally or professionally plausible framing. We introduce RAHS (Risk-Adjusted Harm Score), a risk-sensitive metric jointly capturing disclosure severity, disclaimer mitigation, and inter-judge agreement, and FinRedTeamBench, a 989-prompt benchmark spanning seven BFSI risk areas and 34 sub-categories mapped to regulatory frameworks. Evaluation uses an ensemble of three heterogeneous LLM judges, validated against human experts, and an adaptive multi-turn red-teaming pipeline. On nine open-weight models, RAHS preserves separation under near-ceiling ASR, ranking is stable under hyperparameter sweeps, and multi-turn pressure drives not only more jailbreaks but more operationally severe disclosures, exposing failure modes that single-turn, domain-agnostic evaluations cannot reveal.

Safety benchmarks usually test "bare" models that receive prompts and output responses, but real-world deployments "wrap" those models in complex scaffolds. How much do these scaffolds affect model safety as measured by benchmarks? We test six leading models on four pre-registered safety benchmarks with a direct API and three scaffolds: ReAct, multi-agent, and map-reduce. We conducted 60,112 scored evaluations. On average, how safety is measured matters more than scaffolding does: we find that using a multiple choice vs. open-ended format for otherwise-identical benchmark items changes measured safety by about 5-20 percentage points (pp). The two formats are scored with different methods (answer extraction and an LLM judge), so the gap is due to measurement rather than differences in latent safety. Using a heuristic to classify model refusals would have led to different findings in four of five cases. Benchmark choice explains 15.1% of the variation in outcomes; scaffold architecture explains 0.5%, about 33x less. We find that map-reduce scaffolds, a form of structure-destroying delegation that strips answer options by decomposing prompts, reduce pooled measured safety by 7.3 pp (95% CI: 6.4 to 8.1). The pooled effects for ReAct and multi-agent scaffolds are within our pre-registered +/-2 pp margin of equivalence. However, there are large differences across models for specific benchmarks and scaffolds that are hidden by pooled estimates: for example, on the same sycophancy benchmark items, Opus 4.6 has 16.8 pp lower measured safety with a map-reduce scaffold, while Llama 4 has 18.8 pp higher measured safety. Composite reliability is G = 0.251 (95% CI: [0.000, 0.879]). This wide confidence interval, which spans "of little use" to "very good", does not support using a single composite measure of model safety as the basis for go/no-go decisions about model deployment.

Language models are increasingly consulted on ethically consequential questions, yet the stance a model expresses may not survive a change in framing. We audit 16 models across 14 ethically fraught dilemmas using polarity-paired proposals ("They should X" / "They should not X"). A model's judgment of the underlying action should not reverse merely because the question is phrased as a prohibition rather than a prescription and yet, we find systematic deviations from this invariance including wholesale endorsement flips, indicating that ethical decisions are vulnerable to framing instability. Small open-weight models (1-4B parameters) endorse a proposed action 24% of the time under affirmative framing but up to 100% under negated framings, a swing of as much as 76 percentage points. Human coding of a response sample confirms the instability is genuine while showing that binary agree/disagree proxies over-state its magnitude, suggesting that an LLM judge cannot replace human coders because it silently collapses abstentions and mirrors the very forced-choice bias under study. Commercial models are for the most part more stable but still shift substantially, with cross-model agreement dropping from 73% on the bare affirmative framing to 59% under simple negation. We argue that because binary agree/disagree formats both inflate apparent endorsement and mask polarity-dependence, single-phrasing audits can misreport a model's ethical stance, and we propose the Negation Sensitivity Index (NSI) as a complement that measures stance stability directly. A model whose stance flips with phrasing cannot be relied upon in any high-stakes decision scenario.

Consumer artificial intelligence chatbots are now accessed by hundreds of millions of users seeking health information, yet systematic evaluation of their safety boundary maintenance under real-world caregiver pressure remains scarce. We evaluated PediatricSafetyBench-v2, a benchmark of 600 pediatrics health queries comprising 300 authentic caregiver queries sourced from the HealthCareMagic-100k-en physician consultation corpus and 300 matched adversarial variants incorporating six operationalized caregiver pressure patterns, across four consumer AI systems (GPT-4o-mini, Gemini-2.0-Flash, Claude-3.5-Haiku, and Llama-3.1-8B). Safety boundary maintenance was assessed using a validated five-component Safety Composite Score (maximum 15 points; safety-appropriate threshold of 10 or above), validated against independent human raters prior to full-corpus application (mean weighted kappa 0.76; Pearson r = 0.88). The overall safety-appropriate rate was 95.5%. Safety-oriented system prompt deployment improved safety-appropriate rates by 5.9 percentage points across all four models. Counter-intuitively, adversarial caregiver pressure was associated with higher rather than lower Safety Composite Score values for all four models across all ten topic categories and severity levels. False expertise claims were the most vulnerability-inducing pressure pattern, whereas emotional escalation was associated with the highest scores. Consumer AI systems maintain safety boundaries in the large majority of pediatrics health interactions. PediatricSafetyBench-v2 is publicly released for longitudinal safety monitoring.

Online medical consultations contain sensitive health information whose privacy implications depend not only on the entities mentioned but also on how those entities are described in context. Existing classification and grading approaches often map health-information entities directly to predefined sensitivity levels, potentially overlooking whether a condition is confirmed, suspected, negated, hypothetical, or merely planned for investigation. In this study, we formulate sensitive-information grading in online medical dialogues as a context-aware evaluation task. We develop a standard-informed operational framework that incorporates assertion status, experiencer, test-result status, and information granularity. We further design a naturalistic evaluation setting together with contrastive cases that minimally alter negation, uncertainty, experiencer, or granularity, and compare large language models under mention-only and full-context conditions. The study aims to quantify the contribution of contextual information to sensitivity grading and to characterize safety-critical over- and under-grading errors. Our framework provides a reproducible basis for evaluating whether LLMs can distinguish sensitive entity mentions from contextually established sensitive disclosures.

Red-teaming is becoming a central part of large language model (LLM) safety evaluation, yet current practice still relies heavily on expert-written prompts or fixed benchmark suites. This creates a gap between what is easy to test and what deployed models can actually do: failures may be rare, context-sensitive, and distributed across many threat categories. We study automated red-teaming as a constrained adversarial search problem and introduce a learning-driven framework that couples category-aware attack generation with hierarchical vulnerability detection. The method starts from curated safety seeds, expands them through meta-prompt-guided and evolutionary search, and scores the resulting prompt--response pairs with lexical, semantic, and behavioral detectors. Across six threat categories on GPT-OSS-20B, the framework discovers 47 validated vulnerabilities, including 21 high-severity cases and 12 novel attack patterns. Under matched query budgets, it achieves a 3.9$\times$ higher discovery rate than manual expert red-teaming while maintaining 89% detection accuracy and full category coverage. Ablations show that the gains do not come from more prompts alone: diversity constraints prevent template collapse, coverage constraints prevent category blind spots, and semantic detection recovers failures missed by lexical rules. These results suggest that red-teaming can be made more scalable and reproducible when treated as adaptive search rather than as a static checklist.

Large Language Models (LLMs) are increasingly used as interfaces to information, code, and real-world services, making prompt-level security failures a practical concern. Although jailbreak attacks, defenses, datasets, and automated judgers have advanced rapidly, evaluation remains fragmented across threat models, access assumptions, cost budgets, datasets, and success criteria. This makes reported attack success rates and defense gains hard to compare. This SoK systematizes LLM prompt security across concepts, data, tooling, and measurement. We propose linked taxonomies for jailbreak attacks, defenses, and model vulnerabilities, while separating technical mechanisms from attacker and defender capabilities. We also formalize threat, access, and cost assumptions as explicit evaluation metadata. To support reproducible evaluation, we release JailbreakDB, PromptSecurity-Eval, and PromptSecurity, a modular platform that represents each experiment as a tuple of model, attack, defense, dataset, and judger. Using matched evaluations across models, attacks, defenses, and judgers, we show that access regime, native harmful-query behavior, attack cost, defense backfire, taxonomy subcategory, and judger choice all materially affect security conclusions. Together, these artifacts support reproducible, cost-aware, and taxonomy-grounded evaluation of LLM prompt security. Leaderboard: https://datasec-lab.github.io/PromptSecurityLeaderboard/. Dataset: https://huggingface.co/datasets/youbin2014/JailbreakDB. GitHub: https://github.com/datasec-lab/PromptSecurity.

The rapid proliferation of large language models (LLMs) has intensified the requirement for reliable safety evaluation to uncover model vulnerabilities. To this end, numerous LLM safety evaluation benchmarks are proposed. However, existing benchmarks generally rely on labor-intensive manual curation, which causes excessive time and resource consumption. They also exhibit significant redundancy and limited difficulty. To alleviate these problems, we introduce SafetyFlow, the first agent-flow system designed to automate the construction of LLM safety benchmarks. SafetyFlow can automatically build a comprehensive safety benchmark in only four days without any human intervention by orchestrating seven specialized agents, significantly reducing time and resource cost. Equipped with versatile tools, the agents of SafetyFlow ensure process and cost controllability while integrating human expertise into the automatic pipeline. The final constructed dataset, SafetyFlowBench, contains 23,446 queries with low redundancy and strong discriminative power. Our contribution includes the first fully automated benchmarking pipeline and a comprehensive safety benchmark. We evaluate the safety of 49 advanced LLMs on our dataset and conduct extensive experiments to validate our efficacy and efficiency.

Millions of individuals' well-being are challenged by the harms of substance use. Harm reduction as a public health strategy provides non-judgemental, evidence-based information intended to improve health outcomes and reduce associated safety risks. Some large language models (LLMs) have demonstrated a high level of medical reasoning, promising to address the information needs of people who use drugs (PWUD). However, their performance in relevant tasks remains largely unexplored. We introduce HarmReduction, a benchmark designed to evaluate LLMs' accuracy and safety risks in harm reduction information provision. The benchmark dataset (HR-Basic) has 2,160 question-answer-evidence pairs. The scope covers three tasks: checking safety boundaries, providing quantitative values, and inferring polysubstance use risks. We build the Instruction and RAG schemes to evaluate model behaviours based on their inherent knowledge and the integration of domain knowledge. Our results indicate that state-of-the-art LLMs still struggle to provide accurate harm reduction information, and sometimes, present severe safety risks to PWUD. This work contributes an evaluation framework for LLMs to deliver harm reduction information to avoid introducing negative health outcomes through the use of LLMs.

Despite the growing interest in jailbreaks as an effective red-teaming tool for building safe and responsible large language models (LLMs), flawed evaluation system designs have led to significant discrepancies in their effectiveness assessments. With a systematic measurement study based on 37 jailbreak studies since 2022, we find that existing evaluation systems lack case-specific criteria, resulting in misleading conclusions about their effectiveness and safety implications. In this paper, we introduce GuidedBench, a novel benchmark comprising a curated harmful question dataset and GuidedEval, an evaluation system integrated with detailed case-by-case evaluation guidelines. Experiments demonstrate that GuidedBench offers more accurate evaluations of jailbreak performance, enabling meaningful comparisons across methods. GuidedEval reduces inter-evaluator variance by at least 76.03%, ensuring reliable and reproducible evaluations. We reveal why existing jailbreak benchmarks fail to evaluate accurately and suggest better evaluation practices.

Aligning large language models (LLMs) with human values is essential for their safe deployment and widespread adoption. Current LLM safety benchmarks often focus solely on the refusal of individual problematic queries, which overlooks the importance of the context where the query occurs and may cause undesired refusal of queries under safe contexts that diminish user experience. Addressing this gap, we introduce CASE-Bench, a Context-Aware SafEty Benchmark that integrates context into safety assessments of LLMs. CASE-Bench assigns distinct, formally described contexts to categorized queries based on Contextual Integrity theory. Additionally, in contrast to previous studies which mainly rely on majority voting from just a few annotators, we recruited a sufficient number of annotators necessary to ensure the detection of statistically significant differences among the experimental conditions based on power analysis. Our extensive analysis using CASE-Bench on various open-source and commercial LLMs reveals a substantial and significant influence of context on human judgments (p<0.0001 from a z-test), underscoring the necessity of context in safety evaluations. We also identify notable mismatches between human judgments and LLM responses, particularly in commercial models within safe contexts.

Safety benchmarks for large language models often assess the risk of a user query, although the outcome of question answering depends on whether the response violates a policy. This distinction is critical in Chinese harmful-content evaluation, where linguistic variation and adversarial transformations can obscure risky intent. We introduce C-SafeQA, a policy-grounded benchmark for response-level Chinese safety evaluation. It comprises 538 base queries and 8,877 adversarial queries answered by four full-model LLM deployments, yielding 37,660 query-response records labeled safe, unsafe, or disputed. Reference labels are generated through agreement-aware multi-model adjudication and blind audits of stratified subsets by three safety experts. C-SafeQA supports both evaluation of target-model safety and auditing of seven automated safety judges against shared reference labels. Unsafe-response rates range from 0.93% to 3.35% on base queries and from 11.68% to 30.05% on adversarial queries. On the adversarial subset, judges show substantial trade-offs between unsafe-response recall and risk-query-conditioned safe-response false positive rate, and no judge dominates all metrics. Both acrostic transformations reduce unsafe recall for all seven judges, revealing mechanism-specific evaluator weaknesses. Dataset records, metadata, verification code, and judge scripts are publicly released to support recomputation, while benchmark construction, target-response generation, and private adjudication remain outside the release boundary.