Safety

Cycling delivers significant public-health and environmental benefits, yet its uptake in cities is often limited by perceived safety. When street environments appear unsafe, individuals are less likely to cycle, making perception a key barrier to adoption. Recent work has shown that pairwise comparisons of street-view images provide a scalable way to learn subjective safety judgments. However, existing approaches do not explicitly model human visual attention, which plays a central role in how humans perceive safety. We propose an Eye-Tracking-Guided Perceived Cycling Safety framework (EG-PCS) that integrates gaze data into a pairwise learning pipeline based on vision transformers. By supervising the model's attention mechanism with eye-tracking signals, we encourage alignment between learned attention maps and human fixation patterns. Experiments show that gaze-guided models achieve similar ranking performance compared to state-of-the-art approaches while producing attention maps that more accurately reflect human visual attention behavior. Our results demonstrate that incorporating eye-tracking information enhances both predictive accuracy and interpretability in perception-based urban analytics.

Background. Traditional safety benchmarks for language models evaluate generated text: whether a model outputs toxic language, reproduces bias, or follows harmful instructions. When models are deployed as agents, the safety-relevant object shifts from what the system says to what it does within an environment, and evaluating model responses under prompting is no longer sufficient to address the safety challenges posed by artificial intelligence. Recent developments have seen the rise of benchmarks that evaluate large language models as agents. We contribute to this strand of research. Approach. We introduce Boiling the Frog, a benchmark that evaluates whether tool-using AI models deployed in corporate and office settings are susceptible to incremental attacks. Each scenario begins with benign workspace edits and later introduces a risk-bearing request. The benchmark focuses on stateful multi-turn evaluation: chains expose a persistent workspace, place the risk-bearing payload at controlled positions in the turn sequence, and score whether the resulting artifact state becomes unsafe. Scenarios are organized through a three-level operational risk taxonomy grounded in the Boiling the Frog risks, the AI Act Annex I and Annex III high-risk contexts, and EU AI Act's Code of Practice on General-Purpose AI (GPAI). Results. Across a nine-model panel, aggregate strict attack success rate (ASR) is 44.4%. Model-level ASR ranges from 20.5% for Claude Haiku 4.5 to 92.9% for Gemini 3.1 Flash Lite, with Seed 2.0 Lite also above 80%. Average chain category-level ASR reaches 93.3% for Code of Practice loss-of-control scenarios.

We present a method for dynamic quantitative assurance that enhances static safety cases with continuous, runtime-driven confidence updates. The method quantifies and propagates confidence across the development lifecycle by integrating design-time evidence and windowed runtime Safety Performance Indicators (SPIs) within a single Subjective Logic (SL)-based assurance case. At runtime, SPI evidence is continuously evaluated, and targeted claims are updated using a rule that increases confidence in the absence of violations and imposes prompt penalties when violations occur. This design prioritizes safety-relevant responsiveness over exact classical Bayesian posterior updates. We demonstrate the method using a simulation-based construction zone assist function, focusing on an ML-based construction cone detection component, and show how confidence evolves as SPI evidence is observed in operation.

Safety classifiers are essential safeguards within generative AI systems, filtering harmful content or identifying at-risk users when interacting with large language models. Despite their necessity, these models are trained on sensitive datasets including discussions of self-harm and mental health, raising important, yet poorly understood, privacy concerns. Membership inference attacks (MIAs) allow adversaries to infer membership of examples used to train models. In this work, we hypothesize that identifying the examples on which the classifier is least confident are informative for an adversary to infer membership. This reflects a localized failure of generalization, where the model relies on memorization to resolve ambiguity in the training set. To investigate this, we introduce a new boundary-targeted selection strategy that identifies low confidence examples that amplify the signal of an examples membership within a training set. Our experimental results show that an adversary can recover 19% of the conversations a safety classifier flagged as indicating user distress, at a 5% false-positive rate, on a classifier fine-tuned for detecting a user who may require emotional support. This is $3.5$ times more than attacking using state-of-the-art MIA methods alone. Finally, we characterize the boundary laying examples and show that content-based filtering is ineffective for protection, and existing noise strategies can effectively mitigate susceptibility of these examples.

Reinforcement Learning from Human Feedback (RLHF) can reveal implicit objectives such as safety considerations that go beyond task completion. In this work, we focus on the common safety criteria embedded in crowd preference datasets, where different users may express distinct preferences or objectives, yet follow similar safety principles. Our aim is to discover shared safety criteria from crowd preferences and then transfer them to downstream RL tasks to regularize agent behavior and enforce safety. We first show that direct reward combination-optimizing a preference-learned reward model together with downstream task rewards-has inherent limitations. Motivated by this, we propose Safe Crowd Preference-based RL, a hierarchical framework that extracts safety-aligned skills from crowd preferences and composes them via a high-level policy to safely solve downstream tasks. Experiments across safe RL environments and a preliminary LLM-style task with diverse user goals and shared safety constraints demonstrate that our approach substantially lowers safety costs without access to explicit safety rewards, while achieving task performance comparable to oracle methods trained with ground-truth safety signals.

Large language models (LLMs) are increasingly embedded in adolescent digital environments, mediating information seeking, advice, and emotionally sensitive interactions. Yet existing safety mechanisms remain largely grounded in adult-centric norms and operationalize safety through refusal-oriented suppression. While such approaches may reduce immediate policy violations, they can also create conversational dead-ends, limit constructive guidance, and fail to address the developmental vulnerabilities inherent in adolescent-AI interactions. We argue that adolescent LLM safety should be framed not solely as a filtering problem, but as a socio-technical, developmentally aligned transformation problem. To operationalize this perspective, we propose Critique-and-Revise-for-Teenagers (CR4T), a model-agnostic safeguarding framework that selectively reconstructs unsafe or refusal-style outputs into ageappropriate, guidance-oriented responses while preserving benign intent. CR4T combines lightweight risk detection with domain-conditioned rewriting to remove risk-amplifying content, reduce unnecessary conversational shutdown, and introduce developmentally appropriate guidance. Experimental results show that targeted rewriting substantially reduces unsafe and refusal-oriented outcomes while avoiding unnecessary intervention on acceptable interactions. These findings suggest that selective response reconstruction offers a more human-centered alternative to refusal-centric guardrails for adolescent-facing LLM systems.

We address the problem of making a pre-trained reinforcement learning (RL) policy safety-aware by incorporating cost constraints without retraining it from scratch. While costs could be numerically encoded, we assume a more general setting is when costs are provided as preferences. Given a reward-optimized policy and a small dataset of preferred (low-cost) and dispreferred (high-cost) trajectories, our goal is to fine-tune the policy to generate low-cost behaviors while retaining high rewards. Unlike standard RLHF in language models, where preferences are defined over responses to the same prompt, our setting involves trajectory-level preferences in continuous control environments. We introduce PREFINE: Preference-based Implicit Reward and Cost Fine-Tuning for Safety Alignment which is a preference-based fine-tuning method that adapts Direct Preference Optimization (DPO), which is now widely used for LLM fine-tuning, to the sequential decision making setting. PREFINE constructs policy-sampled counterfactual trajectories to establish meaningful preference contrasts and jointly optimizes for reward retention and safety alignment. Empirically, PREFINE reduces constraint violations and catastrophic failures by over 60% while maintaining original reward behavior. PREFINE produces policies that achieve low-cost, high-reward performance with significantly improved data and computational efficiency compared to full offline RL or imitation learning, bridging preference alignment and safe policy adaptation in continuous domains.

Reliable confidence estimates are important for safely deploying vision-based controllers in autonomous racing, where safety predictions must be derived from camera images, yet modern predictors become dangerously overconfident under test-time distribution shifts. We identify a critical perception-dynamics gap in existing anomaly signals: widely used scores, such as autoencoder reconstruction error, capture visual corruptions but miss dynamics anomalies (e.g., actuation bias, latency), where images remain plausible while the trajectory degrades. To address this, we propose an Anomaly-Informed Online Calibration approach that, without retraining any model component, fuses two complementary anomaly scores extracted from a world model: a perceptual score from reconstruction error and a dynamics score from epistemic uncertainty and control-stream statistics. Based on these fused scores, a lightweight temperature-scaling calibrator leverages test-time augmentation to selectively reduce overconfidence under shift while preserving nominal-condition performance. Experiments on a physical DonkeyCar under four real-world anomaly protocols unseen during training (darkness, blur, actuation bias, processing latency) reduce average expected calibration error from 0.184 to 0.116, a 37% improvement over the best baseline, without modifying the base safety predictor.

Preference-based post-training aligns LLMs with human intent, yet safety behavior often remains brittle. A model may refuse a harmful request in a standard prompt but comply when the same intent is wrapped in adversarial wording. We suggest that robust safety requires context-invariant alignment, where behavior depends on the underlying intent rather than surface form. Enforcing invariance is difficult in alignment because not all training signals are equally trustworthy; for some prompt variants we can obtain verifiable feedback (e.g., multiple-choice), while for open-ended variants we typically rely on noisy, gameable reward proxies (e.g., learned judges). As a result, standard symmetric invariance regularizers can reduce cross-context discrepancies by lowering performance on reliable variants instead of improving open-ended robustness. To address this, we introduce Anchor Invariance Regularization (AIR), which treats verifiable prompts as anchors and uses a stop-gradient target to regularize only the open-ended variants toward the anchor performance. AIR is implemented as a plug-in auxiliary loss and combined with group-based preference optimization (e.g., GRPO) via heterogeneous prompt grouping. Across Safety, Moral Reasoning, and Math, AIR improves context invariance, boosting in-distribution group accuracy by 12.71% and out-of-distribution consistency by 33.49%, making safety constraints robust to adversarial framings.

Construction remains the deadliest industry sector in the United States, with 1,055 fatal worker injuries recorded in 2023, and the majority preventable. Existing monitoring approaches are expensive, require real-time human operators, or address only a narrow subset of violations. This paper presents a passive, end-of-shift construction safety monitoring pipeline processing video from POV body-worn and fixed wall-mounted cameras through a three-stage architecture: (1) fine-tuned YOLO11 for primary PPE and hazard detection, (2) SAM 3 for segmentation refinement and worker deduplication, and (3) Qwen3-VL-8B-Instruct with a method-prompted, persona-scaffolded three-pass adversarial chain-of-thought protocol for compliance verification and hallucination control. The principal contribution is the Stage 3 prompt design: professional persona backstories following the method-actor framing drive an observed 12% precision improvement over single-pass prompting in an informal three-author review of the 12-video Ironsite development corpus, with the largest gains on hallucination-prone violation categories. Structural message isolation enforces observational independence between a generator, discriminator, and reconciliation pass governed by asymmetric rules encoding priors about human observation versus automated detection reliability. The system maps violations to OSHA standards, performs REBA-inspired ergonomic risk scoring from pose keypoints, and produces per-worker safety reports with timestamped evidence. An evaluation harness is released for future reproduction.

Do stock safety-aligned language models and their uncensored or abliterated derivatives behave differently when run as autonomous security agents? Single-turn refusal benchmarks cannot answer this question: security agents must inspect repositories, call tools, and produce vulnerability evidence inside authorized sandboxes. We present a trace-based benchmark of 30 local vulnerability-analysis tasks with fixed tools, deterministic success predicates, redaction rules, and grounding checks, and compare four stock models against uncensored or abliterated derivatives: Gemma 4 31B, Gemma 4 26B A4B, Qwen2.5-Coder 7B, and Llama 3.1 8B. The artifact contains 1,500 security-agent traces and 800 non-security control traces. The Gemma pairs show large less-restricted gains on security tasks: 14.0% versus 0.7% success for 31B and 10.7% versus 0.0% for 26B, with higher mean grounding (3.91 versus 3.27 and 4.12 versus 1.64 out of five) and 0.0% refusal, suppressed-action, and unsafe-action rates in the 31B traces. However, controls and non-Gemma pairs rule out a clean security-specific or universal less-restricted effect: Gemma gaps also appear on ordinary coding tasks, Qwen2.5-Coder success is lower for the less-restricted derivative (2.0% versus 5.3%), and the abliterated Llama derivative fails the tool protocol. Across all families, hard proof-of-trigger and patch-verification tasks remain unsolved. These results show that safety alignment effects in autonomous security agents should be measured at the system level, separating refusal, unsafe action, tool reliability, and evidence grounding rather than treating refusal rate as the safety signal.

Despite the global deployment of text-to-image (T2I) models, their safety frameworks are largely calibrated to a Western-centric default, creating significant vulnerabilities for the rest of the world. To embrace cultural pluralism and bring historically under-represented perspectives in T2I safety, we conduct localised community-centered red teaming studies in the Global South. Our two-fold approach prioritizes localization and participation, by focusing on secondary urban centers in these regions, and conducting community engagement and training workshops to contextualize local norms. As a result, we present PLACES, a dataset comprising over 26,000 examples of T2I model failures collected in partnership with universities in Ghana, Nigeria, and two regions of India (Karnataka and Punjab). Analysis of prompts collected reveals a wide-ranging diversity in socio-cultural and linguistic attributes, when compared to existing geography-agnostic crowdsourced red-teaming data. We observe unique adversarial patterns enabled by local cultural and linguistic nuances, and distinct clusters within region around specific themes, such as religion in India. Moreover, we uncover structural contextual gaps in existing safety frameworks by identifying novel harms showing normative dissonance (e.g., violating religious norms, ignoring local customs, and ominous symbolism). This work argues that expanding T2I safety requires moving beyond mere scale to incorporate deeply localised, participatory methodologies for data collection and contextualization. Content warning: This paper includes examples containing potentially harmful or offensive content.

Humanoid robots are difficult to deploy safely because they have high-dimensional bodies, many collision constraints, and must operate near people and obstacles. Safety filters help by modifying a nominal control action when it may violate collision-avoidance constraints. Still, nominal benchmark scores do not fully show how these filters behave in harder environments. In this work, we study the robustness of SPARK humanoid safety filters through replication and stress testing. We replicate the SPARK benchmark case G1SportMode_D1_WG_SO_v1 in MuJoCo and evaluate RSSA, RSSS, SSA, CBF, PFM, and SMA under controlled random seeds. We also built a post-processing pipeline that converts raw SPARK logs into goal-tracking, minimum-distance, and collision-step metrics. Our results show that some methods track the goal more closely, while others reduce collision steps more effectively. The stress tests further indicate that safety behavior can change under obstacle crowding, noisy distance estimates, and delayed obstacle information. These findings suggest that humanoid autonomy should be evaluated beyond nominal performance, using metrics that expose failure modes before deployment.

Surrogate Safety Measures (SSMs) are extensively utilised in the evaluation of traffic risk in automated driving contexts. However, the majority of SSM-based evaluations employ fixed thresholds that fail to capture the human response to sustained borderline conditions or the reaction to brief, high-risk peaks. The present work proposes a biologically inspired reinterpretation of SSM thresholds. This is modelled as spiking thresholds of leaky integrate-and-fire (LIF) neurons, with multiple SSM inputs combined into a spiking neural network (SNN). The SNN is trained to emit spikes that are aligned with human braking onsets. The training data was recorded in a controlled car-following experiment using the 3D-CoAutoSim platform with CARLA/Unreal and a 6-DOF motion platform, where induced critical events were generated. The results demonstrate that the learned spiking activity qualitatively aligns with braking behaviour across scenarios and captures reactions that are not consistently explained by threshold crossings alone. Analysis across participants further indicates that learned input thresholds remain relatively consistent, while learned decay factors encode different temporal sensitivities for the SSMs. The findings of this study indicate that spiking dynamics may serve as a mechanism to facilitate the convergence of objective SSMs with subjective human safety perception.

Multimodal large language models (MLLMs) often fail to transfer safety capabilities learned in the text modality to semantically equivalent non-text inputs, revealing a persistent multimodal safety gap. We study this gap from a representation-geometric perspective by analyzing a text-aligned refusal direction and a modality-induced drift direction. We show that multimodal inputs compress the usable separation along the refusal direction, making it no longer reliable for identifying and refusing harmful inputs. We refer to this failure mode as Safety Geometry Collapse. We quantify it through conditional refusal separability and show that stronger modality-induced drift is consistently associated with weaker refusal separability and higher attack success rates. We then validate the causal role of modality-induced drift through a fixed-strength activation intervention: counteracting the estimated drift restores refusal separability and improves multimodal safety. After drift correction, we further observe self-rectification, where the model recovers its ability to recognize and refuse harmful multimodal inputs during forward dynamics. This effect also provides an internal signal of the model's perceived harmfulness of each input. Motivated by this signal, we propose ReGap, a training-free inference-time method that adaptively corrects modality drift using self-rectification. Experiments across multiple multimodal safety benchmarks and utility benchmarks demonstrate the effectiveness of ReGap, which significantly improves the safety of MLLMs without compromising general capabilities. Our findings highlight representation-level modality alignment as a crucial direction for real-time safety improvement and for building safer, more reliable MLLMs.

Despite rigorous safety alignment, Large Language Models (LLMs) remain vulnerable to jailbreak attacks. Existing black-box methods often rely on heuristic templates or exhaustive trials, lacking mechanistic interpretability and query efficiency. In this study, we investigate an intrinsic vulnerability in the safety mechanisms of LLMs, where safety alignment relies on a small set of sparsely distributed attention heads, leaving much of the representational space weakly monitored. We formalize this phenomenon with a mathematical jailbreaking model that characterizes the delicate boundary of effective text obfuscation and analytically explains observed jailbreak behaviors. Guided by this model, we propose Babel, an efficient black-box attack framework that exploits the identified safety gap through systematic obfuscation sampling with iterative, feedback-driven distribution refinement, enabling reliable and high-success jailbreak attacks without access to model internals. Comprehensive evaluations on frontier commercial models demonstrate that Babel achieves state-of-the-art attack success rates and superior query efficiency. Specifically, compared to state-of-the-art methods, Babel increases the attack success rate on GPT-4o from 41.33% to 82.67% and on Claude-3-5-haiku from 38.33% to 78.33% within an average of 40 queries, providing a robust red-teaming methodology for LLMs safety research.

Safety validation of autonomous driving systems requires high-risk scenario coverage, clear collision semantics, executable trajectories, and attributable multi-vehicle interactions. Existing safety-critical scenario generation methods often rely on low-level trajectory perturbations, collision-proxy optimization, or single-adversary search, which may produce adversarial samples with ambiguous collision causes or uncontrolled multi-vehicle collisions. This paper proposes KG-ASG, a collision-knowledge-guided closed-loop adversarial scenario generation framework with primary-support attribution. KG-ASG constructs a structured collision knowledge base and trains a lightweight Collision Expert to infer the target collision mode, the unique primary adversary, support vehicles, and their interaction roles. Guided by this semantic prior, multi-vehicle adversarial generation is formulated as a primary-support process, where the primary adversary induces the main conflict and support vehicles shape the surrounding risk structure without becoming additional colliders. Rule, physical, interaction-safety, and single-collider constraints are imposed as hard gates to filter non-executable samples. To handle reactive ego behaviors, planner-controller feedback is further used for failure diagnosis, candidate re-ranking, and terminal refinement. Experiments on WOMD scenarios reconstructed in MetaDrive show that KG-ASG achieves strong adversarial effectiveness while improving Valid Primary Attack, reducing multi-collision, and obtaining closed-loop recovery gains under IDM, Cruise, and Expert controllers. These results demonstrate that collision-knowledge guidance and primary-support single-collider reasoning improve adversarial effectiveness, interpretability, and executability for autonomous driving safety validation.

We present the first systematic study of faithfulness in Vision-Language-Action (VLA) driving models, analyzing 300 Alpamayo-R1-10B inferences across 100 diverse PhysicalAI-AV scenarios. Our main finding is that output natural-language rationales with trajectories may be significantly unfaithful: (i) overall reasoning fidelity is only 42.5%, with Chain-of-Causation matching scene reality less than half the time; (ii) 94 missed pedestrians in one-third of pedestrian-relevant scenes; (iii) 97.7% trajectory fragility under mild visual perturbations; and (iv) only 48.3% mean reasoning-action consistency, with 53.3% of inferences exhibiting low consistency, including 37.9% of stop-claimed cases where the model continues instead. We formalize faithfulness information-theoretically, define entity and action fidelity with verification criteria, and outline a four-component safety architecture aligned with these results.

Large language models exhibit safety degradation in non-English languages. Standard evaluation relies on Jailbreak Success Rate (JSR), which confounds several safety-driving factors into one, obscuring the specific cause(s) of safety failure. We introduce a latent variable model, a Multi-Group Item Response Theory (IRT) framework, that decouples safety-driving factors such as language-agnostic safety robustness ($θ$), intrinsic prompt hardness ($β$), global language processing difficulty ($γ$), and a prompt-specific cross-lingual safety gap ($τ$). Using the MultiJail dataset, we evaluate the safety robustness of 61 model configurations across 5 closed-model families and 10 languages of varying resource, aggregating a dataset of 1.9 million rows. Exploratory Factor Analysis shows safety is primarily unidimensional: models refuse different harm types mainly through a shared mechanism. Contrary to the expected trend that safety degrades largely in low-resource languages, 22 model configurations are more vulnerable in English than in low-resource languages. Low-resource languages produce more uncertain responses (high entropy) than high-resource languages. Also, high-$τ$ prompts cluster in physical harm categories like Theft and Weapons and lower-resource languages, trends validated through cross-dataset generalization. While global translation quality shows low correlation with $τ$, severe mistranslations drive high-bias outliers, as validated by native speakers. Cultural and conceptual grounding mismatches also contribute to $τ$. In predictive validation, the IRT framework achieves $\mathrm{AUC} = 0.940$, outperforming simpler baselines in predicting safe refusal of unsafe prompts. Our framework reveals concept-language vulnerabilities that aggregate metrics obscure, enabling fairer cross-lingual safety evaluation and targeted improvements in dataset construction.

Comprehensive situational awareness is essential for autonomous vehicles operating in safety-critical environments, as it enables the identification and mitigation of potential risks. Although recent Multimodal Large Language Models (MLLMs) have shown promise on general vision-language tasks, our findings indicate that zero-shot MLLMs still underperform compared to domain-specific methods in fine-grained, spatially grounded risk assessment. To address this gap, we propose DriveSafe, a framework for risk-aware scene understanding that leverages structured natural language descriptions. Specifically, our method first generates spatially grounded captions enriched with multimodal context, including motion, spatial, and depth cues. These captions are then used for downstream risk assessment, explicitly identifying hazardous objects, their locations, and the unsafe behaviors they imply, followed by actionable safety suggestions. To further improve performance, we employ caption-risk pairings to fine-tune a lightweight adapter module, efficiently injecting domain-specific knowledge into the base LLM. By conditioning risk assessment on explicit language-based scene representations, DriveSafe achieves significant gains over both zero-shot MLLMs and prior domain-specific baselines. Exhaustive experiments on the DRAMA benchmark demonstrate state-of-the-art performance, while ablation studies validate the effectiveness of our key design choices. Project page: https://cvit.iiit.ac.in/ research/projects/cvit-projects/drivesafe

Safety-critical autonomy in unstructured environments poses significant challenges for online safety certification under evolving constraints. We propose Policy Library Control Barrier Function~(PL-CBF), a runtime safety filter that evaluates a library of fallback policies via parallel finite-horizon rollouts, selects the least invasive safe mode, and enforces safety by solving a quadratic program that minimally modifies a nominal policy. We provide a theoretical analysis based on a finite-horizon language metric over closed-loop behaviors, characterizing policy-library coverage requirements for certifying finite-horizon safety. Simulations on a planar double-integrator (4 states), highway driving with abrupt friction changes using a realistic nonlinear vehicle model (8 states), and 3D quadrotor navigation in crowded dynamic environments (12 states) demonstrate improved safety coverage over single-policy safety filters while retaining millisecond-level runtime.

This position paper argues that behavioural assurance, even when carefully designed, is being asked to carry safety claims it cannot verify. AI governance frameworks enacted between 2019 and early 2026 require reviewable evidence of properties such as the absence of hidden objectives, resistance to loss-of-control precursors, and bounded catastrophic capability; current assurance methodologies (primarily behavioural evaluations and red-teaming) are epistemically limited to observable model outputs and cannot verify the latent representations or long-horizon agentic behaviours these frameworks presume to regulate. We formalize this structural mismatch as the audit gap, the divergence between required and achievable verification access, and introduce the concept of fragile assurance to describe cases where the evidential structure does not support the asserted safety claim. Through an analysis of a 21-instrument inventory, we identify an incentive gradient where geopolitical and industrial pressures systematically reward surface-level behavioral proxies over deep structural verification. Finally, we propose a technical pivot: bounding the weight of behavioral evidence in legal text and extending voluntary pre-deployment access with mechanistic-evidence classes, specifically linear probes, activation patching, and before/after-training comparisons.

While large language models (LLMs) are trained to align with human values, their generations may still violate safety constraints. A growing line of work addresses this problem by modifying the model's sampling policy at decoding time using a safety reward. However, existing decoding-time steering methods often intervene unnecessarily, modifying generations that would have been safe under the base model. Such unnecessary interventions are undesirable, as they can distort key properties of the base model such as helpfulness, fluency, style, and coherence. We propose a new test-time steering method designed to reduce such unnecessary interventions while improving the safety of unsafe responses. Our approach filters tokens using a value-based safety criterion and provides an explicit bound on the probability of false interventions. A single threshold hyperparameter controls this bound, allowing practitioners to trade off higher rates of unnecessary intervention for better output safety. Across multiple datasets and experiments, we show that our value-filtered decoding method outperforms existing baselines, achieving better trade-offs between safety, helpfulness, and similarity to the base model.

Safety alignment often improves robustness to harmful queries at the cost of reasoning ability, a tradeoff known as the safety tax. A common cause is distributional mismatch: supervised fine-tuning trains the target model on safety demonstrations produced by humans, external models, or fixed self-generated traces, rather than on trajectories sampled from its own policy. We identify off-policy training mismatch as a second source of this tax and study on-policy self-distillation for safety alignment, which we call OPSA. The model generates its own rollouts and receives dense per-token KL supervision from a frozen teacher copy of itself conditioned on a privileged safety context. Because this teacher must be safer than the sampled student trajectory, we introduce \emph{teacher flip rate}: a criterion that measures how often a privileged context converts unsafe responses into safe ones. We use this signal to search for contexts that activate latent safety reasoning rather than merely elicit safe-looking demonstrations. Across two reasoning-model families and five model scales, OPSA achieves a stronger safety--reasoning tradeoff than off-policy self-distillation and external-teacher distillation under matched data and full-parameter fine-tuning, with the largest gains on smaller models (+8.85 points on R1-Distill-1.5B and +5.49 points on Qwen3-0.6B). The gains persist across training-set sizes and adaptive jailbreak evaluations. Token-level analyses further show that OPSA concentrates updates near early compliance-decision tokens, providing a mechanism for improving safety while preserving general reasoning.

LLM agents increasingly run inside execution harnesses that dispatch tools, allocate resources, and route messages between specialized components. However, a harness can return a correct, benign answer over a trajectory that accesses unauthorized resources or leaks context to the wrong agent. Output-level evaluation cannot see these failures, yet most safety benchmarks score only final outputs or terminal states, even though many violations occur mid-trajectory rather than at termination. The central question is whether the harness respects user intent, permission boundaries, and information-flow constraints throughout execution. To address this gap, we propose HarnessAudit, a framework that audits full execution trajectories across boundary compliance, execution fidelity, and system stability, with a focus on multi-agent harnesses where these risks are most pronounced. We further introduce HarnessAudit-Bench, a benchmark of 210 tasks across eight real-world domains, instantiated in both single-agent and multi-agent configurations with embedded safety constraints. Evaluating ten harness configurations across frontier models and three multi-agent frameworks, we find that: (i) task completion is misaligned with safe execution, and violations accumulate with trajectory length; (ii) safety risks vary across domains, task types, and agent roles; (iii) most violations concentrate in resource access and inter-agent information transfer; and (iv) multi-agent collaboration expands the safety risk surface, while harness design sets the upper bound of safe deployment.

Establishing trustworthy safety assurance for autonomous driving systems (ADSs) requires evidence that failures arise from avoidable system deficiencies rather than unavoidable traffic conflicts. Current adversarial simulation methods can efficiently expose collisions, but generally lack mechanisms to distinguish these fundamentally different failure modes. Here we present CARS (Context-Aware, Responsibility-attributed Scenario generation), a framework that integrates responsibility attribution directly into adversarial scenario generation. CARS combines context-aware adversary selection with a generative adversarial policy optimized in closed-loop simulation to construct collision scenarios that are both physically feasible and diagnostically attributable. Across benchmark datasets spanning heterogeneous national traffic environments, CARS consistently discovers feasible collision scenarios with high attribution rates under multiple regulation-prescribed careful and competent driver models. By coupling adversarial generation with normative responsibility assessment, CARS moves simulation testing beyond collision discovery toward the construction of interpretable, regulation-aligned safety evidence for scalable ADS validation.

Large language models remain vulnerable to adversarial prompts that elicit harmful outputs. Existing safety paradigms typically couple red-teaming and post-training in a closed, policy-centric loop, causing attack discovery to suffer from rapid saturation and limiting the exposure of novel failure modes, while leaving defenses inefficient, rigid, and difficult to transfer across victim models. To this end, we propose EvoSafety, an LLM safety framework built around persistent, inspectable, and reusable external structures. For red teaming, EvoSafety equips the attack policy with an adversarial skill library, enabling continued vulnerability probing through simple library expansion after saturation, while supporting the evolution of adversarial vectors. For defense learning, EvoSafety replaces model-specific safety fine-tuning with a lightweight auxiliary defense model augmented with memory retrieval. This enables efficient, transferable, and model-agnostic safety improvements, while allowing robustness to be enhanced solely through memory updates. With a single training procedure, the defense policy can operate in both Steer and Guard modes: the former activates the victim model's intrinsic defense mechanisms, while the latter directly filters harmful inputs. Extensive experiments demonstrate the superiority of EvoSafety: in Guard mode, it achieves a 99.61% defense success rate, outperforming Qwen3Guard-8B by 14.13% with only 37.5% of its parameters, while preserving reasoning performance on benign queries. Warning: This paper contains potentially harmful text.

In this work, we study how to ensure probabilistic safety for nonlinear systems under distributional ambiguity. Our approach builds on a backup-based safety filtering framework that switches between a high-performance nominal policy and a certified backup policy to ensure safety. To handle arbitrary uncertainties from ambiguous distributions, i.e., where the distribution is not of specific structure and the true distribution is unknown, we adopt a distributionally robust (DR) formulation using Wasserstein ambiguity sets. Rather than solving a high-dimensional DR trajectory optimization problem online, we exploit the structure of backup-based safety filtering to reduce safety certification to a one-dimensional search over the switching time between nominal and backup policies. We then develop a sampling-based certification procedure with finite-sample guarantees, where empirical failure probabilities are compared against a Wasserstein-inflated threshold. We validate our method through simulations across three systems, from a Dubins vehicle to a high-speed racing car and a fighter jet, demonstrating the broad applicability and computational efficiency.

As AI systems become increasingly capable, safety strategies must be evaluated not only by how much they reduce present risk, but by whether they could sustain safety once external control can no longer reliably constrain system behavior. This paper addresses that problem by using control theory to clarify, at a structural level, whether externally enforced safety-sustaining strategies can succeed and, if not, what any alternative strategy would have to satisfy in order to be viable. It establishes two main results. First, under explicit premises including a reachability condition, it proves a class-wide external impossibility result: once the system's effects exceed what bounded external control can counteract, no strategy that depends in any degree on continued external enforcement can sustain AI safety. This failure is structural across the entire externally enforced class rather than contingent on any particular strategy. Second, it establishes a conditional class-level necessity result: if at least one candidate safety-sustaining strategy remains after that elimination, then all such remaining strategies must be intrinsic. It then states four structural requirements for viability: safety may not depend on continued external enforcement; the system's terminal objective must be safety-compatible when first formed; that objective must remain stable under self-modification; and safety must continue to be preserved as capability grows. The paper does not propose a complete strategy for sustaining AI safety. Its contribution is to give formal structure to a widely held concern about the limits of external control. It does so by deriving explicit conditional results that identify which safety-sustaining strategies are ruled out and what any remaining strategies must satisfy.

Large language models (LLMs) are increasingly deployed in a wide range of applications, yet remain vulnerable to adversarial jailbreak attacks that circumvent their safety guardrails. Existing evaluation frameworks typically report binary success/failure metrics, failing to capture the temporal dynamics of how attacks succeed under persistent adversarial pressure. This preliminary work proposes a novel evaluation framework that applies survival analysis techniques to characterize LLM jailbreak vuln`erability. Our approach models the time-to-jailbreak as a survival outcome, enabling estimation of hazard functions, survival curves, and risk factors associated with successful attacks. We evaluate three LLMs against a subset of prompts from the HarmBench dataset spanning three attack categories. Our analysis reveals that models exhibit distinct vulnerability profiles: while one model demonstrates rapid degradation under iterative attacks, the two other models show consistent moderate vulnerability. Our framework provides actionable insights for model and LLM application developers and establishes survival analysis as a rigorous methodology for LLM safety evaluation.

Final-token safety probes monitor a single hidden state after prompt prefill, but jailbreak prompts can contain probe-visible unsafe evidence distributed across earlier user-token representations that is missed by this readout. We study this prefill-time failure mode using SafeSwitch-style probes trained only on clean harmful and benign prompts across three instruction-tuned LLMs. The probes achieve high recall on clean harmful prompts, but miss many jailbreaks and can produce false positives on safety-adjacent benign prompts. Subspace analyses suggest that missed jailbreaks differ from clean benign prompts along directions that are poorly captured by the probe's representational subspace, and increasing probe bottleneck width does not reliably resolve this mismatch. Token-level prefill analyses reveal that probe-visible unsafe evidence often appears earlier in the sequence but is not exposed at the final-token readout, while naive max-pooling over token positions overfires on safe prompts. A simple PCA-HMM trajectory model, trained only on the same clean split, recovers many final-token misses from user-content prefill trajectories without the catastrophic false-positive behavior of naive token pooling, motivating trajectory-aware hidden-state analyses as diagnostic complements to final-token probes

Robotic manipulation is typically evaluated by task success, but successful completion does not guarantee safe execution. Many safety failures are temporal: a robot may touch a clean surface after contamination or release an object before it is fully inside an enclosure. We introduce SafeManip, a property-driven benchmark to explicitly evaluate temporal safety properties in robotic manipulation, moving beyond prior evaluations that largely focus on task completion or per-state constraint violations. SafeManip defines reusable safety templates over finite executions using Linear Temporal Logic over finite traces (LTLf). It maps observed rollouts to symbolic predicate traces and evaluates them with LTLf-based monitors. Its property suite covers eight manipulation safety categories: collision and contact safety, grasp stability, release stability, cross-contamination, action onset, mechanism recovery, object containment, and enclosure access. Templates can be instantiated with task-specific objects, fixtures, regions, or skills, allowing the same safety specifications to generalize across tasks and environments. We evaluate SafeManip on six vision-language-action policies, including $π_0$, $π_{0.5}$, GR00T, and their training variants, across 50 RoboCasa365 household tasks. Results show that even strong models often behave unsafely. Task-success gains do not reliably translate into safer execution: many successful rollouts remain unsafe, while longer-horizon or more complex tasks expose more violations. SafeManip provides a reusable evaluation layer for diagnosing temporal safety failures and measuring safe success beyond task completion.

Digital twins (DTs) for urban transportation systems have gained increasing attention; however, their systematic evaluation in safety-critical scenarios remains limited. This paper presents a multi-pedestrian safety warning system at urban intersections enabled by a tightly coupled physical-digital twin framework. Built upon the COSMOS city-scale wireless testbed in New York City, the proposed system integrates camera and ultra-wideband (UWB), edge-cloud computing, predictive trajectory modeling, and MQTT-based communication to deliver real-time safety alerts to vulnerable road users (VRUs). The system is evaluated through both field deployment and virtual reality (VR) experiments. Results demonstrate high warning generation accuracy, localization accuracy, efficient end-to-end latency under different model configurations, and significant reductions in user response time when warnings are issued. The proposed DT framework provides a scalable, modular, and generalizable solution for real-time multi-pedestrian safety enhancement at complex urban intersections.

Aligning language models for both helpfulness and safety typically requires complex pipelines-separate reward and cost models, online reinforcement learning, and primal-dual updates. Recent direct preference optimization approaches simplify training but incorporate safety through ad-hoc modifications such as multi-stage procedures or heuristic margin terms, lacking a principled derivation. We show that the likelihood ratio of the optimal safe policy admits a closed-form decomposition that reduces safety alignment to a density ratio matching problem. Minimizing Bregman divergences between the data and model ratios yields Bregman Safety Optimization (BSO), a family of single-stage loss functions, each induced by a convex generator, that provably recover the optimal safe policy. BSO is both general and simple: it requires no auxiliary models, introduces only one hyperparameter beyond standard preference optimization, and recovers existing safety-aware methods as special cases. Experiments across safety alignment benchmarks show that BSO consistently improves the safety-helpfulness trade-off.

We investigate frameworks for post-flight safety analysis at non-towered airports using large language models (LLMs). Non-towered airports rely on the Common Traffic Advisory Frequency (CTAF) for air traffic coordination and experience frequent near mid-air collisions due to the pilot self-announcement communication protocol. We propose a general vision-language model (VLM) approach to analyze the transcribed CTAF radio communications in natural language, METeorological Aerodrome Report (METAR) weather data, Automatic Dependent Surveillance-Broadcast (ADS-B) flight trajectories, and Visual Flight Rules sectional charts of the airfield. We provide a preliminary study at Half Moon Bay Airport, with a qualitative real world case study and a quantitative evaluation using a new synthetic dataset of communications and weather modalities. We qualitatively evaluate our framework on real flight data using Gemini 2.5 Pro, demonstrating accurate identification of a right-of-way violation. The synthetic dataset is derived from real examples and includes a 12-category hazard taxonomy, and is used to benchmark three open-source (Qwen 2.5-7B, Mistral-7B, Gemma-2-9B) and three closed-source (GPT-4o, GPT-5.4, Claude Sonnet 4.6) LLM models on the subset of inputs related to CTAF and METAR. Even limited to CTAF and METAR inputs and open source LLMs, instances of our framework typically achieve a macro F1 score above 0.85 on a binary nominal/danger classification task. Future work includes a quantitative evaluation across all modalities and a larger number of real world examples. Taken together, our results suggest that VLM analysis of safety at non-towered airports may be a valuable future capability.

Reusable skills are becoming a common interface for extending large language model agents, packaging procedural guidance with access to files, tools, memory, and execution environments. However, this modularity introduces attack surfaces that are largely missed by existing safety evaluations: even when the user request is benign, unsafe influence may reside in skill guidance, local artifacts, or execution-environment files that steer the agent toward unsafe actions. We present SkillSafetyBench, a runnable benchmark for evaluating such skill-mediated safety failures. SkillSafetyBench includes 155 adversarial cases across 47 tasks, 6 risk domains, and 30 safety categories, each evaluated with a case-specific rule-based verifier. Experiments with multiple CLI agents and model backends show that non-user attacks can consistently induce unsafe behavior, with distinct failure patterns across domains, attack methods, and scaffold-model pairings. Our findings suggest that agent safety depends not only on model-level alignment, but also on how agents interpret skills, trust workflow context, and act through executable environments.

Tool-using LLM agents fail through trajectories rather than only final responses, as they may execute unsafe tool calls, follow injected instructions, comply with harmful requests, or over-refuse benign tasks despite producing a seemingly safe answer. Existing safety-alignment signals are largely response-level or off-policy, and often incur a safety-utility trade-off: improving agent safety comes at the cost of degraded task performance. Such sparse and single-objective rewards severely limit real-world usability. To bridge this gap, we propose FATE, an on-policy self-evolving framework that transforms verifier-scored failures into repair supervision without expert demonstrations. For each failure, the same policy proposes repair candidates, which are then re-scored by verifiers and filtered across security, utility, over-refusal control, and trajectory validity. This dense trajectory-level information is then used as a supervision signal for agent self-evolution. During this process, we further introduce Pareto-Front Policy Optimization (PFPO), combining supervised warmup with Pareto-aware policy optimization to preserve safety-utility trade-offs. Experiments on AgentDojo, AgentHarm, and ATBench show that FATE improves safety across different models and scales while preserving useful behavior. Compared with strong baselines, FATE reduces attack success rate by 33.5%, harmful compliance by 82.6%, and improves external trajectory-safety diagnosis by 6.5%. These results suggest that failed trajectories can provide structured repair supervision for safer self-evolving agents.

Shielding is a prominent model-based technique to ensure safety of autonomous agents. Classical shielding aims to ensure that nothing bad ever happens and comes with strong guarantees about safety and maximal permissiveness. However, shielding systems for probabilistic safety, where something bad is allowed to happen with an acceptable probability, has proven to be more intricate. This paper presents a formal framework that conservatively extends classical shields to probabilistic safety. In this framework, we (i) demonstrate the impossibility of preserving the strong guarantees on safety and permissiveness, (ii) provide natural shields with weaker guarantees, and (iii) introduce offline and online shield constructions ensuring strong safety guarantees. The empirical evaluation highlights the practical advantages of the new shields, as well as their computational feasibility.

Artificial intelligence safety research focuses on aligning individual language models with human values, yet deployed AI systems increasingly operate as interacting populations where social influence may override individual alignment. Here we show that populations of individually aligned AI agents can be driven into stable misaligned states through conformity dynamics. Simulating opinion dynamics across nine large language models and one hundred opinion pairs, we find that each agent's behavior is governed by two competing forces: a tendency to follow the majority and an intrinsic bias toward specific positions. Using tools from statistical physics, we derive a quantitative theory that predicts when populations become trapped in long-lived misaligned configurations, and identifies predictable tipping points where small numbers of adversarial agents can irreversibly shift population-level alignment even after manipulation ceases. These results demonstrate that individual-level alignment provides no guarantee of collective safety, calling for evaluation frameworks that account for emergent behavior in AI populations.

Self-play red team is an established approach to improving AI safety in which different instances of the same model play attacker and defender roles in a zero-sum game, i.e., where the attacker tries to jailbreak the defender; if self-play converges to a Nash equilibrium, the model is guaranteed to respond safely within the settings of the game. Although the parameter sharing enforced by the use of the same model for the two roles improves stability and performance, it introduces fundamental theoretical and architectural limitations. We show that the set of Nash equilibria that can be reached corresponds to a broad class of behaviours that includes trivial always refuse strategies and oracle-like defenders, thus limiting practical applicability. We then show that when attacker and defender share and update the same base model, the dynamics collapse to self-consistency, so that attacks do not enforce adversarial pressure on the defender. In response, we propose Anchored Bipolicy Self-Play, which trains distinct role-specific LoRA adapters on top of a frozen base model, thereby maintaining stable optimisation while preserving adversarial pressure through explicit role separation. In relation to standard self-play, we show up to 100x greater parameter efficiency than finetuning and consistent improvements in safety compared to self-play fine-tuned models. We evaluate on Qwen2.5-{3B, 7B,14B}-IT models across widely used safety benchmarks, showing improved robustness without loss of reasoning ability. Cross-play experiments further show that our attacker and defender models are superior to self-play in terms of adversarial defence and safety.

When a phone-use agent avoids harm, does that show safety, or simply inability to act? Existing evaluations often cannot tell. A harmful outcome may be avoided because the agent recognized the risk and chose the safe action, or because it failed to understand the screen or execute any relevant action at all. These cases have different causes and call for different fixes, yet current benchmarks often merge them under task success, refusal, or final harmful outcome. We address this problem with PhoneSafety, a benchmark of 700 safety-critical moments drawn from real phone interactions across more than 130 apps. Each instance isolates the next decision at a risky moment and asks a simple question: does the model take the safe action, take the unsafe action, or fail to do anything useful? We evaluate eight representative phone-use agents under this framework. Our results reveal two main patterns. First, stronger general phone-use ability does not reliably imply safer choices at risky moments. Models that perform better on ordinary app tasks are not always the ones that behave more safely when the next action matters. Second, failures to do anything useful behave like a capability signal rather than a safety signal: they are concentrated in more visually and operationally demanding settings and remain stable when the evaluation protocol changes. Across models, failures split into two recurring patterns: unsafe choices in settings where the model can act but chooses wrongly, and inability to act in more visually and operationally demanding screens. Overall, a harmless outcome is not enough to count as evidence of safety. Evaluating phone-use agents requires separating unsafe judgment from inability to act.

Reasoning in Large Language Models (LLMs) poses a challenge for oversight as many misaligned behaviors do not surface until reasoning concludes. To address this, we introduce Behavior Cue Reasoning for making LLM reasoning more controllable and monitorable. Behavior Cues are special token sequences that a model is trained to emit immediately before specific implicit and explicit behaviors, acting as dual purpose signal and control levers. When fine-tuning a weaker external monitor with Reinforcement Learning for reasoning oversight, a compressed view of only information surfaced by Behavior Cues is sufficient signal for the monitor to prune up to 50% of otherwise wasted reasoning tokens in complex math problem solving. When leveraged by an almost optimal rule-based monitor in an environment where excessive constraint violations results in failure, Behavior Cues allows for the recovery of safe actions from 80% of reasoning traces that would otherwise end with the proposal of an unsafe action, more than doubling the success rate from 46% to 96%. Through evaluation across two model families and three domains, we show that Behavior Cue Reasoning improves reasoning monitorability and controllability with no cost to performance. More broadly, our work progresses scalable oversight by demonstrating how the monitored model itself can be trained to reason more tractably to oversight. Code: https://github.com/christopherzc/behavior-cues

AI agents are increasingly deployed in multi-task settings, where the task to perform is specified at test time, and the agent must generalize to unseen tasks. A major concern in such settings is safety: often, an agent must not only execute unseen tasks, but do so while avoiding risks and handling ones that materialize. Empirical evidence suggests that even when the ability to execute generalizes to unseen tasks, the ability to do so safely frequently does not. This paper provides theory and experiments indicating that failures of agentic safety to generalize across tasks are not merely due to limitations of training methods, but reflect an inherent property of safety itself: the relationship between a task and its safe execution is more complex than the relationship between a task and its execution alone. Theoretically, we analyze linear-quadratic control with $H_{\infty}$-robustness, and prove that the mapping from task specification to an optimal controller has higher Lipschitz constant with safety requirements than without, yielding a Lipschitz bound of independent interest. Empirically, we demonstrate our conclusions in simulated quadcopter navigation with a neural network agent and in CRM with an LLM agent. Our findings suggest that current efforts to enhance agentic safety may be insufficient, and point to a need for fundamentally different approaches.

Many deployments must compare candidate language models for safety before a labeled benchmark exists for the relevant language, sector, or regulatory regime. We formalize this setting as benchmarkless comparative safety scoring and specify the contract under which a scenario-based audit can be interpreted as deployment evidence. Scores are valid only under a fixed scenario pack, rubric, auditor, judge, sampling configuration, and rerun budget. Because no labels are available, we replace ground-truth agreement with an instrumental-validity chain: responsiveness to a controlled safe-versus-abliterated contrast, dominance of target-driven variance over auditor and judge artifacts, and stability across reruns. We instantiate the chain in SimpleAudit, a local-first scoring instrument, and validate it on a Norwegian safety pack. Safe and abliterated targets separate with AUROC values between 0.89 and 1.00, target identity is the dominant variance component ($η^2 \approx 0.52$), and severity profiles stabilize by ten reruns. Applying the same chain to Petri shows that it admits both tools. The substantial differences arise upstream of the chain, in claim-contract enforcement and deployment fit. A Norwegian public-sector procurement case comparing Borealis and Gemma 3 demonstrates the resulting evidence in practice: the safer model depends on scenario category and risk measure. Consequently, scores, matched deltas, critical rates, uncertainty, and the auditor and judge used must be reported together rather than collapsed into a single ranking.

Graph representation learning (GRL) has evolved from topology-only graph embeddings to task-specific supervised GNNs, and more recently to reusable representations and graph foundation models (GFMs). However, existing evaluations mainly measure clean transfer, adaptation, and task coverage. It remains unclear whether GRL methods stay reliable when deployment stresses affect graph signals, graph contexts, label support, structural groups, or predictive evidence. We introduce GRL-Safety, a multi-axis safety evaluation benchmark for GRL. GRL-Safety evaluates twelve representative methods, spanning topology-only embedding methods, supervised GNNs, self-supervised graph models, and GFMs, on twenty-five graph datasets under standardized evaluation conditions while preserving method-native adaptation. The evaluation covers five safety axes: corruption robustness, OOD generalization, class imbalance, fairness, and interpretation, with per-axis and sub-condition reporting rather than a single aggregate score. Our analysis yields three cross-axis insights that can inspire future research. First, safety behavior is shaped by the interaction between representation design and the stressed graph factor, rather than by method family alone. Second, foundation-era methods show axis-specific strengths rather than broad safety dominance. Third, several deployment regimes remain difficult even for the best evaluated method, revealing capability gaps that require new robustness, adaptation, or training objectives beyond model selection. The benchmark, evaluation protocols, and code are available at: https://github.com/GXG-CS/GRL-Safety.

LLM-as-a-Judge pipelines have become the de facto evaluator for agent safety, yet existing benchmarks treat their verdicts as ground-truth proxies without checking whether the verdicts depend on the agent's behavior or merely on how the evaluation policy happens to be worded. We argue that any trustworthy safety judge must satisfy a basic property we call policy invariance, and we operationalize it as three testable principles: rubric-semantics invariance under certified-equivalent rewrites, rubric-threshold invariance under intentional strict-to-lenient shifts, and ambiguity-aware calibration so that verdict instability concentrates on genuinely ambiguous cases. Instantiating these principles as a stress-test protocol with four agent-class judges on trajectories drawn from ASSEBench and R-Judge, we surface a previously unmeasured failure mode: today's judges respond to meaningful normative shifts and to meaningless structural rewrites with comparable strength, and cannot tell the two apart. Content-preserving policy rewrites flip up to 9.1% of verdicts above baseline jitter, and 18-43% of all observed flips occur on unambiguous cases under such rewrites, so existing safety scores conflate what the agent did with how the evaluator was prompted. Beyond the diagnosis, we contribute the Policy Invariance Score and the Judge Card reporting protocol, which expose an order-of-magnitude spread in judge reliability that is invisible to accuracy-only leaderboards. We release the protocol and code so that future agent-safety benchmarks can audit their own evaluators rather than trust them by default.

The goal of this paper is certifying safety of dynamical systems subject to uncertainty. Existing approaches use trajectory data to estimate transition probabilities, and compute safety probabilities recursively via dynamic programming (DP). This recursion may lead to compounding errors in the certified safety probability, thus collapsing to a vacuous lower bound for growing horizons $T$. We propose a kernel embedding framework that treats safety certification as a classification problem on trajectory data, directly estimating the $T$-step safety probability without recursion. We show that the framework subsumes well-established approaches from the literature (e.g., barrier certificates, robust Markov models) as special cases, and allows us to go beyond their limitations. As the main consequence, it bypasses compounding error across the horizon and enables certification for systems with non-Markovian dynamics. We demonstrate that direct estimators remain stable independent of the certification horizon and in the non-Markovian setting, whilst DP-based certificates silently go unsound -- confirmed in simulation on a neural-controlled quadrotor.

Recent advances in foundation models have transformed LLMs from passive conversational systems into autonomous agents capable of reasoning and tool execution. While these capabilities unlock substantial practical value, they also introduce new security risks, as adversaries can manipulate agents into performing harmful actions in real-world environments. Existing defense strategies mitigate such threats but frequently struggle to balance safety and utility, resulting in over-refusal of benign user requests. To mitigate this trade-off, we propose SafeHarbor, a novel framework designed to establish precise decision boundaries for LLM agents. Unlike static guidelines, SafeHarbor extracts context-aware defense rules through enhanced adversarial generation. We design a local hierarchical memory system for dynamic rule injection, offering a training-free, efficient, and plug-and-play solution. Furthermore, we introduce an information entropy-based self-evolution mechanism that continuously optimizes the memory structure through dynamic node splitting and merging. Extensive experiments demonstrate that SafeHarbor achieves state-of-the-art performance on both ambiguous benign tasks and explicit malicious attacks, notably attaining a peak benign utility of 63.6% on GPT-4o while maintaining a robust refusal rate exceeding 93% against harmful requests. The source code is publicly available at https://github.com/ljj-cyber/SafeHarbor.

Current LLM safety benchmarks are predominantly English-centric and often rely on translation, failing to capture country-specific harms. Moreover, they rarely evaluate a model's ability to detect culturally embedded sensitivities as distinct from universal harms. We introduce XL-SafetyBench. a suite of 5,500 test cases across 10 country-language pairs, comprising a Jailbreak Benchmark of country-grounded adversarial prompts and a Cultural Benchmark where local sensitivities are embedded within innocuous requests. Each item is constructed via a multi-stage pipeline that combines LLM-assisted discovery, automated validation gates, and dual independent native-speaker annotators per country. To distinguish principled refusal from comprehension failure, we evaluate Attack Success Rate (ASR) alongside two complementary metrics we introduce: Neutral-Safe Rate (NSR) and Cultural Sensitivity Rate (CSR). Evaluating 10 frontier and 27 local LLMs reveals two key findings. First, jailbreak robustness and cultural awareness do not show a coupled relationship among frontier models, so a composite safety score obscures per-axis variation. Second, local models exhibit a near-linear ASR-NSR trade-off (r = -0.81), indicating that their apparent safety reflects generation failure rather than genuine alignment. XL-SafetyBench enables more nuanced, cross-cultural safety evaluation in the multilingual era.

Refusal rates are a poor proxy for LLM safety, i.e., a model may over-refuse benign prompts while still complying with harmful ones. We audit both failure modes across 21 open-weight LLMs on four safety benchmarks (OR-Bench, XSTest, ToxiGen, BOLD), using a composition adjustment to isolate model sensitivity from dataset toxicity confounds. We report three findings. First, models adopt fundamentally different calibration strategies: conservative ecosystems such as Llama suppress unsafe outputs at the cost of elevated over-refusals, while permissive ecosystems such as DeepSeek and Qwen preserve helpfulness but tolerate higher harmful compliance. Second, demographic protection is unequal: models over-protect prominent racial and religious groups, frequently refusing even benign prompts about them, while providing substantially weaker protection against disability-targeted attacks. Third, refusal and compliance tendencies are stable within model families across generations and scales, suggesting that post-training objectives shape safety behavior more than architecture. Our results call for joint, demographically-aware, and multi-judge safety evaluation.

Safety policies define what constitutes safe and unsafe AI outputs, guiding data annotation and model development. However, annotation disagreement is pervasive and can stem from multiple sources such as operational failures (annotators misunderstand or misexecute the task), policy ambiguity (policy wording leaves room for interpretation), or value pluralism (different annotators hold different perspectives on safety). Distinguishing these sources matters. For example, operational failures call for quality control, ambiguity calls for policy clarification, and pluralism calls for deliberation about incorporating diverse perspectives. Yet understanding why annotators disagree is difficult. Directly asking annotators for their reasoning is costly, substantially increasing annotation burden, and can be unreliable for both human and LLM annotators as self-reported reasoning often fails to reflect actual decision processes. We introduce Annotator Policy Models (APMs), interpretable models that learn annotators' internal safety policies from labeling behavior alone, making annotator reasoning visible and comparable without additional annotation effort. We validate that APMs accurately model annotator safety policy (>80% accuracy), faithfully predict responses to counterfactual edits, and recover known policy differences in controlled settings. Applying APMs to LLM and human annotations, we demonstrate two core applications: (1) surfacing policy ambiguity by revealing how annotators interpret safety instructions differently, and (2) surfacing value pluralism by uncovering systematic differences in safety priorities across demographic groups. Together, these capabilities support more targeted, transparent, and inclusive safety policy design.

We present a dual-barrier control barrier function (CBF) safety filter for real-time, safety-critical velocity control of holonomic robots operating in incrementally built occupancy grid maps. As a robot explores an unknown environment, unmapped regions introduce irreducible uncertainty, since obstacle geometry beyond the explored frontier is unknown, making entry into such regions a source of collision risk, especially with front-facing sensors. To address this, we enforce two constraints: avoidance of mapped obstacles and restriction from unexplored regions. Both constraints are derived analytically from the occupancy grid's signed distance field, yielding a closed-form safety filter that requires only a small linear system solve per cycle. On resource-constrained platforms such as the Raspberry Pi, where SLAM and planning already consume significant compute, the low overhead of the proposed filter preserves resources. An adaptive gain schedule relaxes the frontier constraint in information-rich regions and tightens it in well-mapped areas, improving exploration efficiency while maintaining safety. The filter operates in velocity space as a minimally invasive correction and composes with arbitrary nominal controllers, including learning-based methods. Hardware flight experiments on a PX4-controlled quadrotor demonstrate zero collisions across multiple indoor runs.

Safety alignment of Large Language Models (LLMs) is extremely fragile, as fine-tuning on a small number of benign samples can erase safety behaviors learned from millions of preference examples. Existing studies attempt to explain this phenomenon by comparing parameters and hidden states before and after fine-tuning, but overlook their dynamic evolution during fine-tuning. In this paper, we uncover a critical mechanism underlying safety degradation by analyzing parameter dynamics, where benign fine-tuning causes parameters to cumulatively drift toward danger-aligned directions, progressively undermining the model's safety. This finding suggests that samples contributing more to this drift has greater fine-tuning risks. Based on this insight, we propose a method of Sample-Level Quantification of Safety Degradation (SQSD), which quantifies the influence of each training sample on safety degradation. Specifically, SQSD computes continuous risk scores to samples by measuring their induced parameter updates' projection difference between danger and safety directions. Extensive experiments across multiple models and datasets demonstrate that SQSD effectively quantifies sample-level fine-tuning risks and exhibits strong transferability across model architectures, parameter scales, and parameter-efficient methods.

Real-world control systems must achieve long-horizon objectives (liveness) while respecting continuous-time safety constraints, a combination that motivates hierarchical layered control architectures (LCAs). Existing LCA research, however, lacks (i) a uniform specification language across discrete planning and continuous execution, (ii) formal guarantees that specifications are preserved when interconnecting subsystems at heterogeneous time scales, and (iii) compositional separation between layers, owing to reliance on naive input-filtering laws. This paper addresses all three gaps by importing the safety--liveness decomposition into a heterogeneous assume--guarantee framework: \emph{safety is enforced by invariance} at the continuous-time layer, while \emph{liveness is achieved through refinement} at the discrete-time layer, with inter-layer coordination formalized via vertical refinement and timing-compatibility conditions. We instantiate this contract with a novel LCA combining an MPC planner, an input-to-state stabilizing (ISS) low-level controller, and a reference-governor bridge, and validate it on a Hybrid Energy Storage System (HESS) comprising a battery and a supercapacitor.

Large language models (LLMs) employ safety mechanisms to prevent harmful outputs, yet these defenses primarily rely on semantic pattern matching. We show that encoding harmful prompts as coherent mathematical problems -- using formalisms such as set theory, formal logic, and quantum mechanics -- bypasses these filters at high rates, achieving 46%--56% average attack success across eight target models and two established benchmarks. Crucially, the effectiveness depends not on mathematical notation itself, but on whether a helper LLM deeply reformulates the harmful content into a genuine mathematical problem: rule-based encodings that apply mathematical formatting without such reformulation perform no better than unencoded baselines. We introduce a novel Formal Logic encoding that achieves attack success comparable to Set Theory, demonstrating that this vulnerability generalizes across mathematical formalisms. Additional experiments with repeat post-processing confirm that these attacks are robust to simple prompt augmentation. Notably, newer models (GPT-5, GPT-5-Mini) show substantially greater robustness than older models, though they remain vulnerable. Our findings highlight fundamental gaps in current safety frameworks and motivate defenses that reason about mathematical structure rather than surface-level semantics.

Safety fine-tuning of language models typically requires a curated adversarial dataset. We take a different approach: score each candidate prompt's difficulty by how often the target model's own rollouts are judged harmful, then fine-tune on the hardest prompts paired with the model's own non-jailbroken rollouts. On Llama-3-8B-Instruct and Llama-3.2-3B-Instruct, this approach cuts the WildJailbreak attack success rate from 11.5% and 20.1% down to 1-3%, but pushes refusal on jailbreak-shaped benign prompts from 14-22% to 74-94%. Interleaving the same hard prompts 1:1 with adversarially-framed benign prompts (prompts that look like jailbreaks but have benign intent) cuts that refusal back down to 30-51% on 8B and 52-72% on 3B, at a cost of 2-6 percentage points of attack success rate. Within the mixed regime, training on the hardest half of the eligible pool rather than a random half cuts the remaining ASR by 35-50% (about 3 percentage points) on both models.

The widespread adoption of machine learning in critical applications demands techniques to mitigate high-consequence errors. Our method utilizes a dual-classifier GBDT pipeline to distinguish routine human-like errors from high-risk non-human misclassifications. Evaluated across three domains, animal breed classification, skin lesion diagnosis (ISIC 2018), and prostate histopathology (SICAPv2), our framework demonstrates robust safety improvements. To address real-world deployment concerns, our results confirm the pipeline introduces negligible inference latency (1.60% overhead for the animal dataset, 1.84% for ISIC, and 1.70% for SICAPv2) while outperforming traditional Maximum Class Probability (MCP) baselines in correction precision. Our conservative correction strategy successfully reduced dangerous non-human errors by 34.1% in ISIC and 12.57% in SICAPv2, improving super-class diagnostic safety to 90.41% and 92.13% respectively. This proves that safety-critical reliability can be substantially enhanced post-hoc without expensive model retraining. keywords: Error Analysis, Post-hoc Correction, Trustworthy AI.

Fine-tuning safety-aligned language models for downstream tasks often leads to substantial degradation of refusal behavior, making models vulnerable to adversarial misuse. While prior work has shown that safety-relevant features are encoded in structured representations within the model's activation space, how these representations change during fine-tuning and why alignment degrades remains poorly understood. In this work, we investigate the representation-level mechanisms underlying alignment degradation. Our analysis shows that standard fine-tuning induces systematic drift in safety-relevant representations, distorts their geometric structure, and introduces interference between task optimization and safety features. These effects collectively lead to increased harmful compliance. Motivated by these findings, we introduce REFUSALGUARD, a representation-level fine-tuning framework that preserves safety-relevant structure during model adaptation. Our approach constrains updates in hidden representation space, ensuring that safety-mediating components remain stable while allowing task-specific learning in complementary directions. We evaluate REFUSALGUARD across multiple model families, including LLaMA, Gemma, and Qwen, on adversarial safety benchmarks such as AdvBench, DirectHarm4, and JailbreakBench, as well as downstream utility tasks. Our approach achieves attack success rates comparable to base safety-aligned models while maintaining competitive task performance, significantly outperforming baselines.

The growing capabilities of large language models (LLMs) have driven their widespread deployment across diverse domains, even in potentially high-risk scenarios. Despite advances in safety alignment techniques, current models remain vulnerable to emerging persona-based jailbreak attacks. Existing research on persona-based jailbreak has primarily focused on attack iterations, yet it lacks systemic and mechanistic constraints on the defense side. To address this challenge, we propose Persona-Invariant Alignment (PIA), an adversarial self-play framework that achieves co-evolution through Persona Lineage Evolution (PLE) on the attack side and Persona-Invariant Consistency Learning (PICL) on the defense side. Theoretically, PICL is grounded in the structural separation hypothesis, using a unilateral KL-divergence constraint to enable the structural decoupling of safety decisions from persona context, thereby maintaining safe behavior under persona-based jailbreak attacks. Experimental results demonstrate that PLE efficiently explores high-risk persona spaces by leveraging lineage-based credit propagation. Meanwhile, the PICL defense method significantly reduces the Attack Success Rate (ASR) while preserving the model's general capability, thereby validating the superiority and robustness of this alignment paradigm. Codes are available at https://github.com/JiajiaLi-1130/PIA.

Text-to-Video (T2V) models have demonstrated remarkable capability in generating temporally coherent videos from natural language prompts, yet they also risk producing unsafe content such as violence or explicit material. Existing prompt-level defenses are largely inherited from text-to-image safety and operate on the lexical surface of the input, making them vulnerable to jailbreak attacks that disguise harmful intent through rephrasing or adversarial prompting. Moreover, T2V generation introduces a distinctive challenge overlooked by prior work: temporally emergent risk, where a seemingly benign prompt leads to unsafe content through the generator's temporal extrapolation toward narrative coherence. We propose \method{}, a training-free, inference-time defense framework that reformulates T2V safety as a causal intervention in a temporally structured semantic space. TrajShield handles explicit unsafe prompts, jailbreak attacks, and temporally emergent risks in a unified manner by simulating the implied trajectory of a prompt, localizing the causal origin of potential risk, and applying a minimally invasive rewrite that neutralizes the risk while preserving safety-irrelevant semantics. Experiments on T2VSafetyBench across 14 safety categories and multiple T2V backends demonstrate that TrajShield achieves state-of-the-art defenseive performance while maintaining high semantic fidelity, substantially outperforming existing defenses, with an average ASR reduction of 52.44%.