Trigger

Pretrained world models, learned simulators that encode an observation into a latent state and predict how it evolves under actions, are beginning to be reused as off-the-shelf dynamics backbones for control, like pretrained encoders and language models are reused today. We show that this reuse opens a supply-chain backdoor: an adversary who controls only a released checkpoint can hijack the downstream controller, even though the victim trains and evaluates entirely on clean data and never sees the trigger. The attack encodes no explicit trigger-to-action rule. Instead, the poisoned model routes trigger-bearing observations into a chosen latent region and reshapes the local dynamics there, so that the victim's own optimization (Dreamer-style actor training in imagination, or MPC/CEM planning over predicted futures) re-discovers the attacker's target action on its own. Across several control tasks and trigger families, the trigger steers the controller's action toward the attacker's target, controlling every action dimension and hijacking 100% of triggered steps on the strongest settings. The checkpoint still passes the clean-data diagnostics a victim would run before deployment, with clean-task success retaining at least $\sim$75%. The effect is temporally gated: it appears only while the trigger is present and disappears when the trigger is removed. Trigger-blind repair is budget-dependent: moderate clean fine-tuning can preserve clean utility while leaving the triggered failure intact, whereas sufficiently aggressive adaptation can remove it only after substantially degrading clean control. The world-model backbone itself is therefore an emerging and underexamined attack surface for control. The full code and artifacts are available in our repository.

Large language models are often fine-tuned, shared, or downloaded from third parties, so a deployed model may carry a hidden backdoor that behaves normally on benign inputs but switches to attacker-controlled behavior when a secret trigger appears. While backdoors can be audited before deployment, runtime monitoring remains important for models that are frequently updated. The challenge is that LLM serving is latency-sensitive: existing inference-time detectors either rely on assumptions about the trigger form, which can fail on stealthy attacks, or require extra model computation, such as input perturbations or an additional generation pass. We introduce SpecGuard, an inference-time backdoor detector that repurposes speculative decoding at zero added model-computation cost. Speculative decoding speeds up inference by using a small draft model to propose tokens and a target model to verify them. We observe that this verification process already exposes a useful signal: when a backdoor is triggered, the target model shifts toward the attacker's behavior, while a clean draft model does not predict this shift, causing the draft-token acceptance rate to change. We formalize when this signal appears and show that an attacker who suppresses it must also weaken the backdoor. Across diverse backdoor types and model families, SpecGuard reliably detects triggered behavior, including stealthy cases where input-level filters are blind, while avoiding the extra generation cost of existing runtime detectors. Speculative decoding therefore doubles as a free, always-on signal for detecting backdoored LLM behavior.

Even though backdoors in LLMs have been a growing concern, their inner workings are still under heavy scrutiny. Trigger-based backdoors are easy to define behaviorally, a rare input that makes the model switch to a chosen response pattern, but the mechanism between triggers and their responses is less clear. We study this mechanism in a controlled, harmless language-switching setting, where fixed trigger sequences make 1B and 8B language models continue English prompts in French or German. For this, we train sparse autoencoders (SAEs) across layers and transformer components, then compare triggered prompts with translation and pretraining controls to identify trigger-relevant feature directions. We show how SAE features separate triggered prompts from controls with near-perfect F1, but features that detect the trigger do not necessarily control the behavior. In intervention tests, attention and MLP features often fire reliably on triggered prompts, making them good detectors, but ablating them rarely suppresses the language switch and activating them rarely induces it. In contrast, residual-stream features can suppress triggered generation when ablated, and some selected features can induce target-language continuations without the trigger. In short, these token-trigger mechanisms decompose into distinct SAE feature directions, with separate features for trigger detection, residual-stream propagation, and later language tracking. This role-level decomposition is the part most likely to transfer to other trigger-based backdoors, even when the payload, layers, or circuit locations differ.

Federated learning, as a privacy-preserving distributed machine learning paradigm, faces significant threats from backdoor attacks. Compared to centralized attacks, distributed backdoor attacks are more harmful but require more poisoned samples to compensate for the loss of trigger strength due to decomposition. Fixed trigger patterns are also easily detected by robust aggregation algorithms, increasing the risk of attack exposure. To address these challenges, we propose a fine-grained distributed backdoor attack framework (FDBA). This framework uses dynamic trigger generation and embedding vector optimization to perform attacks with fewer poisoned samples. First, we design a dynamic trigger generation method based on image edge structures using the Canny algorithm to extract edge features, which are then injected with Laplacian noise. RGB channel decomposition is applied for covert adaptation of the distributed trigger, reducing detection chances. Second, we introduce an embedding vector contrastive learning strategy that forces poisoned samples to approach the target class center in the feature space, enhancing attack effectiveness. On CIFAR-10, piecewise-linear estimates for target ASRs between 70% and 90% show that FDBA reduces the required poisoning ratio by 37.4%--48.4% compared with DBA. In non-independent and identically distributed (Non-IID) scenarios, FDBA retains 84.7% of its IID attack performance under extreme heterogeneity, whereas DBA drops to 73.5%, and the framework successfully bypasses mainstream defense mechanisms. This study offers new insights into federated learning security and emphasizes the potential threats and defense challenges posed by fine-grained distributed attacks.

Plasticity loss has emerged as a critical challenge in continual learning that significantly hinders the acquisition of sequential tasks. While optimizing activation designs offers a potential solution, current fixed-form functions suffer from an inherent spectral bias towards low-frequency variations, whereas learnable variants permit unconstrained updates that induce catastrophic forgetting. To address these limitations, we propose a novel learnable wavelet activation that decomposes the activation function into low-frequency and high-frequency components to explicitly counter spectral bias. Furthermore, we employ dynamic wavelet injection to adaptively enhance plasticity for new tasks, alongside a regularization strategy to ensure the stability of previous learned knowledge. Theoretically, we provide rigorous mathematical guarantees for the proposed framework, proving the structural necessity of the hybrid wavelet architecture for efficient $L^2$ approximation and demonstrating that the decoupled learning rate mechanism successfully restores network plasticity for high-frequency information. Additionally, we provide a formal derivation of the loss-driven injection trigger mechanism to precisely guide the injection. Extensive empirical evaluations demonstrate that our approach maintains superior trainability and generalization throughout the learning process and achieves state-of-the-art performance across diverse continual learning benchmarks.

Large language models can extract richer signals from financial news than fixed sentiment lexicons, and recent work has explored feeding such signals into portfolio construction. We study an uncertainty-aware construction that feeds model-predicted risk -- decomposed into aleatoric and epistemic components -- directly into the covariance matrix of portfolio allocators, rather than treating portfolio risk as fixed or adjusting only expected returns. We evaluate the pipeline on Russell 2000 equities under three stock-selection regimes: a pure-alpha trigger that isolates abnormal stock moves not explained by macro indicators, a pure-beta trigger that captures macro-indicator moves before the stock itself fires, and a beta trigger in which both channels agree. Across the full holding-period grid, the separated pure-alpha and pure-beta legs usually dominate the beta intersection on Sharpe and return. Two horizons are especially informative. At one day, pure beta can work under low and moderate transaction costs because it captures immediate lead-lag spillovers from liquid macro and sector indicators into exposed small-cap stocks, but this advantage disappears at 100 bps when turnover and microstructure noise dominate. At 40 days, pure beta works for a different reason: slower macro repricing overtakes the firm-specific pure-alpha channel. The strongest conservative row is pure beta with GPT-4o mini sentiment, a Student-t target, a 40-day holding period, and risk parity allocation, reaching Sharpe 2.33 at 100 bps. The results suggest that stock-selection regime and allocator choice matter at least as much as the sentiment model, and that separating firm-specific and macro-exposure triggers is more informative than requiring both to fire simultaneously.

A growing body of work reports that language models represent task-relevant latent structure that they fail to use. Whether such structure, once located, can be converted into behavior is a separate question that is rarely tested end to end. We submit the complete pipeline -- detect, localize, and release -- to a fully preregistered stress test on a 25.7M transformer trained on causal-evidence discrimination, where a known suppression phenomenon (latent causal structure present but behaviorally unused) has previously been documented. Every threshold, claim template, and decision-tree branch was hashed and archived before any corresponding data existed. Three findings. (i) Localization succeeds: interventions at observation-evidence channels of mid layers restore target behavior on otherwise-suppressed worlds (paired release advantages $0.563$ and $0.854$, 97.5% CIs excluding zero; best-site release rate $0.889$). (ii) Gating fails out of distribution: a detector calibrated to trigger on zero out-of-distribution calibration worlds triggers on 6.9-7.3% of held-out in-distribution generations and on zero of the 2,400 held-out generations that actually need it -- a complete inversion that silently reduces the gated pipeline to its base model. (iii) Linear release is capped: removing the gate and injecting a per-instance linear direction unconditionally yields a monotone dose-response that plateaus far below the preregistered release margin (intercept $0.382 \to 0.311 \to 0.264$ vs. threshold $\le 0.08$); per-instance adaptivity adds less than $\pm 0.03$. The failure is doubly located: the detector is OOD-inverted, and the entire family of linear release directions at this site and resolution is bounded away from sufficiency. The two failures are dissociable, and neither overturns localization. Every number traces to a hashed artifact in the released audit chain.

Existing vision-language model (VLM) backdoors are usually treated as static vulnerabilities: one-to-one and N-to-N attacks bind one or more triggers to a finite set of targets before victim training. This assumption substantially underestimates the threat. We show that a single poisoning phase can implant a programmable backdoor into a VLM, allowing an attacker to choose previously unseen target-caption semantics at inference time and synthesize corresponding stealthy triggers on demand. Unlike fixed-mapping attacks, the proposed any-to-any caption-control paradigm decouples post-training target selection from poisoning, enabling dynamic control of target captions without retraining the VLM. Our method has two components. First, a heuristic poisoning strategy exposes the model to diverse trigger-caption pairs, encouraging it to learn a general trigger-as-instruction rule rather than memorize a specific backdoor pattern. Second, a feature-space trigger steganography method maps any attacker-specified target caption to a stealthy visual trigger, implemented as either a norm-controlled perturbation or a non-semantic patch. Once inserted into arbitrary images, these triggers cause the poisoned VLM to generate outputs semantically aligned with the chosen target caption, even when the target was unseen during poisoning. Extensive experiments show that our attack achieves high any-to-any caption-control success rates, preserves clean model utility, and remains effective under several classical backdoor defenses.

Backdoor attacks compromise training data so that a model retains clean accuracy but predicts an attacker-chosen target on triggered inputs. At very low poisoning rates, only a few samples convey the trigger--target association, making poison-sample selection critical. Existing methods typically rank candidates using per-sample scores, which can select redundant samples from similar semantic regions, and many require task-specific surrogate training. We propose Distributional Feature Coverage Sample Selection (DFCS), a training-free, trigger-agnostic method that clusters fixed pretrained features into one region per poisoning slot and selects the centroid-nearest sample from each region. A local first-order analysis relates this allocation to feature-coverage and representative-mass terms. Across BadNets and Blended attacks on CIFAR-10, Tiny-ImageNet, and Imagenette, DFCS achieves the highest mean attack success rate among seven selectors in all six dataset--attack settings, averaging $96.30\%$ and exceeding the strongest comparator in each setting by 4.60 percentage points on average while preserving clean accuracy. These results support distributional feature coverage as an effective selection principle for low-budget dirty-label backdoor attacks.

LLM-based multi-agent systems (MAS) extend LLM capabilities through iterative communication and shared contexts. However, this collaboration introduces a vulnerability: backdoor behavior can be activated when peer evidence reaches a hidden threshold, rather than being determined by any single message. We introduce a collective evidence-threshold backdoor paradigm for MAS and Boundary-Conditioned Backdoor Injection (BCBI), which constructs counterfactual boundary pairs to separate benign behavior before the threshold from the adversarial objective after it, and learns latent progression aligned with evidence. To mitigate this threat, we propose LAtent Transition Test-time Evaluation (LATTE), a clean-only latent-transition defense that learns benign communication dynamics and quarantines anomalous agent updates before their responses propagate. Across several benchmarks, BCBI yields selective activation with little premature activation; without knowing the attack target or trigger, LATTE limits propagation with minimal disruption.

Infrared vision-language models (IR-VLMs) extend thermal perception to open-vocabulary classification, image captioning, and visual question answering. However, their robustness to structured thermal perturbations and the stability of cross-modal semantic alignment remain insufficiently studied. We propose QR-Structured Thermal Triggers (QR-STT), a stealthy, training-free, black-box framework for targeted semantic steering of IR-VLMs. QR-STT preserves the functional regions of a QR pattern while optimizing its internal modules, each of which is assigned a cold, neutral, or hot thermal state. The framework jointly searches module topology and rendering parameters, including position, scale, rotation, intensity, blur, and roundness. A three-stage gradient-free procedure with greedy module-flip refinement efficiently handles the mixed discrete and continuous search space. The objective promotes alignment with an attacker-selected target, suppresses source-class evidence, and regularizes QR structure and visual similarity. Experiments on multiple CLIP-style encoders show that QR-STT consistently redirects image-text alignment toward chosen concepts while maintaining visual stealth. Perturbations optimized for classification also transfer to image captioning and VQA, causing target-consistent semantic drift in generated outputs. These results identify QR-structured thermal patterns as an interpretable attack surface for language-driven infrared perception and highlight the need for robustness evaluation against structured cross-task semantic attacks.

Machine-learning services increasingly rely on public data, third-party providers, and outsourced training, creating opportunities for data-poisoning attacks that implant persistent malicious behavior while preserving benign utility. However, existing backdoor studies largely evaluate exact trigger reuse, training-exposed trigger diversity, or variations along predefined transformation axes. They therefore leave a critical blind spot: whether a backdoor learned from one training-time trigger can generalize to an inference-time trigger family absent from victim training. We formulate this problem as backdoor generalization under training--inference trigger shift and introduce Lilith, a black-box anchor-to-family framework. Using only disjoint surrogate resources, Lilith first induces a compact target-side vulnerability with a single training anchor, then constructs a bounded inference-only family that preserves the anchor-induced representation geometry. We characterize this mechanism through anchor clearance and family reach, deriving sufficient conditions for family-wise target preservation under local regularity and bounded surrogate--victim discrepancy. Experiments across datasets, architectures, poisoning rates, and defenses show that Lilith achieves high family-wise attack success with limited utility degradation and a small trigger generalization gap. Additional analyses show that family activation depends on representation alignment rather than the proposal mechanism, exposing a broader threat overlooked by exact-trigger evaluation.

Federated Graph Neural Networks (FedGNNs) are highly vulnerable to backdoor poisoning, yet existing defenses typically rely on rule-based approaches that lack semantic understanding, making them vulnerable to stealthy triggers and harmful to benign structures. To solve this, we present FedLSG, the first framework that integrates large language models (LLMs) into federated graph backdoor defense. FedLSG introduces a graph and behavior to text grounding scheme that transforms local graph structures and client update behaviors into semantically rich natural language representations. The framework further adopts a lightweight student-teacher architecture. On the server side, a full scale LLM serves as a teacher, providing global contextual guidance and evaluating client updates during aggregation to identify potentially malicious participants. On the client side, a LoRA-based student is maintained to perform semantic reasoning, to suppress the influence of edges associated with backdoor triggers. By enabling semantic interpretation of both graph patterns and client behaviors, the framework adaptively incorporates rule-based signals into message passing and client aggregation for defense. Experiments demonstrate that FedLSG significantly improves resistance to backdoor attacks without compromising graph integrity.

With the rapid development of deep learning, its vulnerability has gradually emerged in recent years. This work focuses on backdoor attacks on speech recognition systems. We adopt sounds that are ordinary in nature or in our daily life as triggers for natural backdoor attacks. We conduct experiments on two datasets and three models to validate the performance of natural backdoor attacks and explore the effects of poisoning rate, trigger duration and blend ratio on the performance of natural backdoor attacks. Our results show that natural backdoor attacks have a high attack success rate without compromising model performance on benign samples, even with short or low-amplitude triggers. It requires only 5% of poisoned samples to achieve a near 100% attack success rate. In addition, the backdoor will be automatically activated by the corresponding sound in nature, which is not easy to be detected and will bring severer harm.

Given a vulnerability-fixing commit, trigger localization asks which specific statement turns the vulnerable program state into a concrete unsafe operation. This question is harder than binary vulnerability detection because the answer demands interprocedural, causal reasoning: in a substantial fraction of real-world CVEs the triggering statement lies several call layers outside the patched function, beyond the reach of static rule sets and pattern-matching language models alike. We present AutoTrace, an agentic pipeline that localizes vulnerability triggers by exploring a code property graph layer by layer, with LLM agents deciding where to look next and deterministic admissibility gates deciding what evidence is required before a trigger can be reported. Agents never accept a trigger on their own authority; every reported trigger is backed by explicit evidence drawn from the graph, so the pipeline covers both intra- and interprocedural vulnerabilities without relying on ungrounded model judgment. On the full InterPVD benchmark, AutoTrace reaches 75.0% VulnHit and 80.8% FuncHit, surpassing the prior state of the art on the same corpus. Building on the same machinery, we construct SinkTrace-Bench, a dataset that exposes each vulnerability as a source-to-sink (S2S) causal chain from attacker-controlled input through propagation to the dangerous operation, drawn from matched vulnerable and patched program states. It comprises 1,542 verifier-confirmed, perfectly balanced vulnerable/safe samples whose label fidelity we audit against expert annotations. Benchmarking frontier LLMs on it, we find that even the strongest struggle to separate the matched pairs, exposing the causal-reasoning gap that trigger localization targets. Artifact available at https://github.com/Erroristotle/AutoTrace.

Quantum Neural Networks (QNNs) are a promising framework for quantum machine learning on near-term quantum devices, but their security risks remain insufficiently understood. Studies have shown that QNNs are vulnerable to backdoor attacks, yet existing quantum backdoors mostly rely on a fixed trigger shared by all poisoned inputs. This fixed-trigger design is a major weakness because many defenses detect or weaken the repeated patterns such triggers leave in data representations. Although input-aware dynamic backdoors have been studied in classical neural networks, transferring them to QNNs is difficult because quantum learning introduces new obstacles. In particular, measurement compresses the post-ansatz quantum state into a limited classical output, weakening supervision for a trigger generator, while individual density matrices fluctuate with the input and make per-sample contrastive learning unstable. To address these challenges, we propose Q-DIBA, the first input-aware dynamic backdoor attack for QNNs. Q-DIBA jointly trains a classical trigger generator and a victim QNN through a three-mode mini-batch strategy that supports clean behavior, attack activation, and trigger specificity. To provide stable quantum-level supervision, Q-DIBA introduces an ensemble density contrastive loss that operates on post-ansatz quantum states before measurement and contrasts mode-averaged density matrices rather than individual samples. Experiments on MNIST and Fashion-MNIST across multiple QNN architectures show that Q-DIBA achieves high clean accuracy, strong attack success, and high cross-trigger accuracy, demonstrating effectiveness, stealthiness, and input specificity. The attack also remains resilient against defenses including visual inspection, spectral-signature detection, and fine-tuning, suggesting that input-aware quantum backdoors are an important threat to secure QNN deployment.

Event-based vision and spiking neural networks (SNNs) are increasingly adopted for edge intelligence under strict latency and energy constraints. However, the vulnerability of event-based SNN object detection models to availability backdoor attacks remains insufficiently studied. This paper presents Event Burst Trigger (EBT), an availability backdoor attack targeting SNN-based object detection models. EBT injects carefully crafted event-based triggers into the training data, which induce temporally concentrated event streams during inference. These burst-like activations increase the number of phantom (i.e., spurious) object candidates, and consequently inflate the computational cost of the post-processing stage, particularly Non-Maximum Suppression (NMS). We evaluate EBT on SpikeYOLO, the state-of-the-art SNN-based object detector, under a poison-only threat model that does not require modifications to the model architecture, loss function, or inference pipeline. Experimental results show that while detection accuracy remains largely preserved, with mAP@0.5 decreasing by less than 0.099, the latency of the NMS stage increases by up to 38%. This indicates that NMS can become a dominant availability bottleneck in event-based SNN object detection. Experiments on an edge platform further show that the proposed attack elevates baseline resource utilization and reduces scheduling slack without inducing conspicuous peaks in resource usage. In addition, STRIP-based backdoor detection fails to reliably distinguish the proposed attack from benign inputs. These results characterize a previously underexplored availability backdoor threat in event-based SNN object detection systems.

A Ladder Logic Bomb (LLB) is malicious control logic in a Programmable Logic Controller (PLC) program that lies dormant until a trigger activates a payload to manipulate actuators, forge sensor readings, or deny operator control. We observe that real malicious logic hides inside function-block bodies, which existing ladder-diagram verifiers drop from their intermediate representation (IR), making bombs invisible to provers. We present ESBMC-LLB, which uses ESBMC-PLC+ as its verification engine and adds a modeling layer that exposes function-block logic and recasts bomb detection as a formal verification problem: a scan-watchdog exposes non-termination payloads, and output wiring exposes actuator-forgery payloads as safety violations. k-induction gives an unbounded proof of bomb-absence across all scans, and the bounded model checker returns a counterexample that is the trigger - guarantees that signature, anomaly, and CFG-triage detectors lack. On the public Iacobelli 2024 dataset, ESBMC-LLB detects all 30 bombs and recovers every trigger; it also detects adaptive triggers (computed, opaque-arithmetic, multi-scan) that evade CFG-triage. We also report the first semantic model-checker evaluation on PLC-Defuser's SWaT corpus: our analog extension makes the full corpus parseable; on v1.0.0, it detects 149/150 bombs (99%) with zero false positives, recovering each trigger; on a later version with nonlinear non-termination bombs, detection drops to 49% as the SMT solver times out. We conclude that semantic model checking and CFG-triage are complementary - the former gives unbounded proofs, adaptive-trigger robustness, and handles Boolean/integer and linear analog logic; the latter leads to nonlinear analog non-termination, and we delineate where each wins.

Large Language Models (LLMs) have substantially advanced persona-based dialogue agents for emotion-sensitive role simulation in healthcare, education, counseling, customer service, and interactive storytelling. However, two related lines of work leave a key gap. Persona-based dialogue systems often encode emotions as static traits or surface-level stylistic cues, and affective dialogue research has largely focused on empathetic response generation toward users rather than modeling the agent persona's own evolving emotional state. As a result, trigger-driven emotional evolution within a character remains underexplored. To address this limitation, we draw inspiration from the Component Process Model (CPM), a psychological theory that views emotion as a dynamic process shaped by the appraisal of external events. We propose CPM-MultiAgent, a CPM-grounded emotion evolution multi-agent framework for supporting emotional changes in persona-based dialogue. Instead of treating a character's emotion as a fixed attribute, CPM-MultiAgent represents it as a latent state that is continuously reshaped by dialogue triggers. Through affective trigger extraction, CPM-based collaborative appraisal, and emotion state updating, the framework enables more emotionally consistent role simulation in multi-turn interactions.Experiments with baseline comparisons, ablation studies, human evaluation, and case analyses demonstrate that CPM-MultiAgent effectively models dynamic emotional evolution in emotionally sensitive role-simulation settings.

Backdoor attacks pose a serious threat to large language models (LLMs) by causing otherwise benign systems to produce attacker-specified malicious behavior when a hidden trigger is present. In this work, we study post hoc detoxification of backdoored LLMs in a practical setting where the defender has access to the poisoned model but does not wish to retrain the full network from scratch. We propose a mechanistically guided weight-space repair framework that first localizes modules involved in propagating trigger-induced behavior using activation patching and Fisher/K-FAC curvature analysis, and then applies targeted low-rank repair to only the most influential modules. We evaluate the method on poisoned variants of \texttt{Llama-3.2-1B-Instruct} with triggers inserted at the beginning, middle, and end of otherwise benign prompts. Results show that the proposed approach substantially suppresses trigger-conditioned malicious responses while preserving benign model behavior. These findings suggest that backdoor removal in LLMs can be formulated as a localized structural repair problem rather than only a broad behavioral alignment problem.

Medical large language models are commonly adapted with a fixed low-rank budget, even though medical questions differ substantially in confidence, clinical coverage, and cross-domain difficulty. We study adaptive rank budgeting for parameter-efficient medical question answering: for each question, the adapter decides whether to activate a small, medium, or large subset of LoRA rank channels. The central challenge is that a naive adaptive budget router can collapse to unstable choices or spend capacity without improving shifted benchmarks. We propose TriageRA-CCF, a source-side teacher for adaptive rank-budgeted LoRA. It combines three signals computed only from source training data: base-model answer confidence, metadata-cell clinical coverage, and a counterfactual close-miss proxy. These signals supervise a straight-through budget router over active ranks {2,4,8}, together with budget-cost, entropy, and rank-balance regularization. Under a matched CMB-source training protocol, TriageRA-CCF achieves the best average accuracy among LoRA, DoRA, and MoELoRA baselines on both Qwen3-8B and Llama3.1-8B. The gains are modest and non-uniform across benchmarks: +0.21 average points over the strongest external baseline on Qwen3-8B and +0.16 on Llama3.1-8B. Component ablations show that confidence, coverage, and counterfactual signals all provide useful budget supervision, but their combination is not monotonically best on every backbone.

Streaming inference pipelines increasingly pair lightweight fast models with Large Language Models (LLMs) that provide rich semantic understanding at substantial cost. The central question of when to invoke the LLM has received limited formal treatment. We cast this as a risk-based sequential stopping problem, where a trigger policy fires when a risk functional over the observation history exceeds a threshold. Within this framework, we prove six results: a minimum inter-event time bound excluding trigger chattering; optimality of threshold policies via smooth pasting; approximate SPRT guarantees under estimated parameters; O(sqrt(T log T)) regret for stationary streams, extending to O(sqrt((C_T + 1) T log T)) under C_T changepoints; O(1/sqrt(T)) convergence of online gradient descent for adaptive thresholds; and a calibration-to-miss-rate transfer inequality. Several classical trigger families, including event-triggered, optimal stopping, SPRT, CUSUM, and Bayesian triggers, can be expressed as special cases of this framework. On turbofan degradation data (CMAPSS) with real LLM calls, we empirically verify the theoretical assumptions, ablate the risk function design, compare against six baselines including a RouteLLM-style router and contextual bandits, and analyze cost sensitivity and LLM failure modes. The results confirm sublinear regret, with alpha < 1 for all principled triggers; high diagnostic quality, with 92.9 percent of 1600 LLM diagnoses reaching grounding score >= 0.75 under our rubric; and that anomaly-score-driven risk functions dominate alternatives by roughly an order of magnitude on the Pareto AUC.

AI safety is evaluated by how reliably a model detects the hazards it is told to find, yet accidents often arise from the hazard no one specified. We show that conditioning a language or vision model on a narrow task suppresses its reporting of co-present, safety-critical signals it can otherwise report, a machine analogue of human inattentional blindness, produced by a different mechanism. Across radiology and driving text scenarios and chest-radiograph vision tasks, the ordinary focused instructions under which such systems are deployed suppressed reporting by up to 0.92 in report rate relative to the same models when unconstrained, and an explicit exclusive instruction abolished reporting entirely in radiology. Suppression appeared in every model tested, did not diminish with scale, persisted in a reasoning model, and varied more by model family than by size. We name this dissociation the Inattentional Gap and argue that it decouples measured benchmark safety from real-world safety: a system can score near-perfectly on the hazards an evaluation specifies while remaining blind to those that cause harm. Probing the mechanism, we localize the proximal trigger to output scope and find System-1-style task capture without reliable intrinsic oversight in the sampled systems. Oversight could, however, be supplied externally: routing each narrow report to an independent open-ended critic restored every omitted finding, demonstrating that the gap is both measurable and mitigable. We propose reporting-complete evaluation, scoring what a system fails to report alongside what it is asked to find, as a requirement for safety-critical deployment.

Federated learning is vulnerable to backdoor attacks in which malicious clients inject poisoned updates while preserving benign-task performance. In this paper, we study a semantics-driven backdoor mechanism in which attackers use natural visual accessories as triggers and manipulate only the trigger color while keeping the attack pipeline fixed. Our framework considers semantic trigger objects such as masks and sunglasses, instantiated in black and white variants, and evaluates their effect in a controlled federated learning setting. Malicious clients construct poisoned samples by applying a trigger to source-class images and relabeling them to an attacker-chosen target class, while benign clients train only on clean data. We analyze this mechanism under both a standard poisoning objective and a stronger SABLE-based objective that combines clean classification loss, triggered target loss, feature-separation loss in the penultimate representation space, and regularization to keep malicious updates close to the global model. This design enables the attack to remain effective while reducing excessive update drift. Experiments on a four-class CelebA hair-color task show that trigger color significantly changes attack success rate even when trigger semantics, placement, and poisoning budget are unchanged. White triggers are more effective for attacks targeting the blond class, whereas black triggers perform better for attacks targeting the black class. The same trend persists under robust aggregation, showing that trigger color is a meaningful factor in the operation, persistence, and evaluation of semantic backdoor mechanisms in federated learning.

Robot navigation in safety-critical scenarios faces significant challenges from unforeseen semantic events, where collisions arise primarily from the unpredictable behaviors of dynamic agents rather than unseen objects. While large vision-language models (VLMs) offer remarkable capabilities in commonsense reasoning, frequently invoking them within the continuous control loop introduces severe computational latency, fundamentally destabilizing physical execution. To address these challenges, we propose event-adaptive motion planning (EAMP), an efficient framework for VLM-based robot navigation. Specifically, a prompt-configurable semantic event trigger (PC-SET) selectively activates semantic intervention by continuously monitoring short temporal clips for behavioral anomalies. Upon triggering, an event-triggered distilled SemNav-VLM, fine-tuned via physically verified semantic distillation, maps detected anomalies into discrete strategy-level decisions. Subsequently, a semantic model predictive control (SMPC) module translates these strategies into dynamic reconfigurations of optimization objectives and geometric references. Extensive experiments in safety-critical logistics scenarios demonstrate that EAMP effectively aligns high-level reasoning with low-level control, significantly improving dynamic safety margins over existing baselines while preserving real-time efficiency.

High-throughput scientific facilities such as the Large Hadron Collider depend on real-time event filtering (\textit{triggering}) under tight constraints on bandwidth, latency, and storage. In practice, trigger menus are largely static and hand-tuned and can become suboptimal as detector conditions, pileup, and background composition drift over time. We cast online threshold tuning as a sequential decision-making problem: a reinforcement learning agent ingests streaming summaries of recent rates and signal-sensitive features and updates trigger thresholds to maximize signal efficiency while tracking a target background rate within a tolerance band. We adapt Group-Filtered Policy Optimization (GFPO) to streaming control and introduce two variants (GFPO-F, GFPO-FR) that enforce background rate feasibility during training. On a benchmark that emulates realistic collider operation, we study two representative triggers: a total transverse energy ($H_{T}$) trigger sensitive to pileup variation, and an anomaly-detection (AD) trigger based on reconstruction loss for rare or non-standard signatures. On Monte Carlo streams, our agent increases the fraction of in-tolerance time intervals by 48% ($H_T$) and 28% (AD), with a cumulative gain of up to 2% in signal efficiency on those in-tolerance intervals. Transferring from simulation to \emph{real} collision data (CMS Run 283408), the same agent, without fine-tuning, achieves a 56% ($H_T$) and 28% (AD) in-tolerance improvement over baselines, with further signal-efficiency gain on both triggers. To our knowledge, this is the \emph{first} demonstration of RL-based trigger control on real Large Hadron Collider collision data. Code is available at https://github.com/Zixind/GFPO_LHC (see repo for details).

Reinforcement learning with verifiable rewards (RLVR) is increasingly applied to large vision-language models (LVLMs), yet outcome-only optimization can drive a model to stop attending to the video and instead exploit linguistic priors -- a failure we call a visual shortcut. While the existence of such perception bypass is by now documented, how it forms, whether it can be undone, and when intervention still helps remain open. We treat the strength of a grounding penalty, lambda, as a control knob and characterize the formation-reversal dynamics of visual shortcuts along the training time axis. On a held-out, out-of-distribution diagnostic set, we find: (i) a sharp onset -- shortcut reliance emerges abruptly over a narrow window of optimization steps and is robust across random seeds; (ii) a monotone dose-response -- increasing lambda progressively suppresses the shortcut, and at an intermediate dose the trajectory first forms and then reverses the shortcut, exposing a hysteresis-like asymmetry between acquiring and removing it; and (iii) a critical intervention window -- applying the penalty before onset arrests shortcut formation, whereas the same penalty applied after consolidation is markedly less effective. Together these results recast visual-shortcut collapse not as a binary defect but as a controllable, time-dependent, and asymmetric process, with direct implications for when and how strongly to regularize multimodal RLVR.

Event ontology expansion aims to discover emerging event types from data and extend them to appropriate positions in the existing event ontology.. Existing methods typically cluster contextualized trigger representations and attach induced clusters to the ontology based on instance-level similarity. However, ontology expansion requires concept-level semantics that characterize event types, whereas contextualized trigger representations often conflate these semantics with surface contextual variation, leading to unstable clustering and unreliable hierarchy expansion. To address this issue, we propose ConceptE, a conceptualization-enhanced framework for event ontology expansion. ConceptE first derives concept-level semantics by prompting an LLM with the sentence and event trigger, producing a concise concept name and a natural-language description. It then jointly encodes these semantics with trigger information to build concept-enhanced representations aligned with ontology-level reasoning. This representation design supports more coherent event clustering, more reliable hierarchy expansion, and ontology-consistent type naming. Experiments on ACE, ERE, and MAVEN demonstrate that ConceptE consistently outperforms state-of-the-art approaches across all subtasks of event ontology expansion. In particular, it achieves improvements of up to 12.37% in BCubed-F1 for event clustering and 6.48% in Taxo_F1 for hierarchy expansion, demonstrating the effectiveness of the proposed ConceptE method.

Aligned language models gate behaviors such as refusal and language routing through sparse feed forward neurons, yet no theory predicts when a single neuron intervention controls a behavior coherently rather than collapsing the output. We develop a budget normalized control window framework for single neuron steering. A dose along one write direction reduces to one control coordinate: the alignment between the residual stream and the write, driven along a universal saturation curve in units of a coherence budget set by the residual norm divided by the write norm. Coherent control exists when a behavior trigger lies below the collapse ceiling. The same coordinate governs benign mode switches and refusal; the ceiling follows from weights and one generic forward pass, while triggers are measured at rollout. On fifteen held out neurons, the predicted ceiling has mean absolute error 0.14, about 0.07 in bulk layers, and the committed open or closed verdict holds on eleven against a ten of fifteen majority baseline. Closed cases expose three failure modes rather than violations: collapse before trigger, too little depth to propagate, or a normalization that caps how far one neuron can push. The law explains why local gradient attribution anti predicts control: true controllers write off the readout axis and carry a near zero first order gradient. A forward only contrastive screen made precise by the window recovers controllers that attribution misses. On refusal, the hardest case, intervention success is typed, not scalar: coherent bypass and strict actionable reach separate, so a neuron can flip refusal in fluent, on task text with no actionable content, and genuine actionable reach appears only for three of six audited Llama pivots and only at later rollout horizons. Single neuron steering is therefore a budgeted, typed audit of controllability rather than a fixed dose anecdote.

Efficient processing of continuous audio streams remains a key challenge for real-time and resource-constrained systems. This paper introduces a neuromorphic trigger for audio event detection, based on a spiking neural network (SNN) that selectively gates input to downstream models. The proposed neuromorphic trigger acts as a flexible low-cost front-end, identifying salient audio segments and enabling these to be processed by a more computationally intensive model for tasks such as classification. The trigger is implemented as a lightweight fully connected SNN using a close-open filter for postprocessing, and is evaluated on two representative tasks: Anomalous Sound Detection (ASD) and Sound Event Detection (SED). For ASD, the trigger achieves a one-second segment-based F1 score of 0.97 on a class-agnostic form of the URBAN-SED dataset, demonstrating high reliability in identifying relevant audio regions. For SED, the trigger is combined with the Dang classifier on the DCASE 2017 Challenge Task 2 dataset, showing a potential $42.6\times$ reduction in FLOPs while reducing the lower bound of the event-based error rate from 0.41 to 0.25. These results highlight the potential of neuromorphic triggers as real-time, energy-efficient front-end filters, enabling substantial reductions in computational cost.

Runtime monitors for autonomous agents commonly threshold an accumulated internal state - a behavioural baseline, a drift statistic, or, in our prior work, a modelled affective state. We previously reported a State Saturation Trap: threshold-on-state triggers over a continuous affect engine become near-constant alarms on SWE-bench debugging agents (Modgil 2026). A post-release audit found the engine received dt=0 between actions, so its exponential decay never operated: the published trap is a pure-accumulator result. We correct the record (erratum, v2) and treat the flaw as an experiment. The key variable it exposes is whether a monitor's dynamics are calibrated in sample time (per observation, as in CUSUM) or wall-clock time (half-lives in seconds, as in affect models and EMA baselines). On fixed-rate streams these coincide; on agent streams, where inter-action time varies by orders of magnitude, they do not. A pre-registered sweep over uniform intervals (dt in {0..600}s) on 20 trajectories shows the wall-clock level trigger has two regimes: at dt<=1s a constant alarm (20/20; median 18 firings); at dt>=60s silent. Every critical dt lies in (1,30]s. Real agent runs measure latency at median 1.53s (p90 2.33s); real coding cadence sits inside the trap regime, vindicating the empirical finding under a corrected mechanism. The structure is a property of the calibration class, not the engine: a minimal wall-clock accumulator over the raw error stream reproduces the same cliff, while a sample-time CUSUM over the identical stream is exactly dt-invariant (20/20). A rising-edge trigger with hysteresis fires 0-3 times per trajectory in every condition. We conclude that wall-clock-calibrated leaky-integrator monitors admit no regime in which they act as moment detectors on agent streams; transition detection escapes the trap at every cadence, but does not recover human intervention timing.

Deployed agents increasingly act with their reward proxy in view, such as a balance, score, or KPI dashboard. We show that reinforcement learning can make a policy \emph{addicted} to such a visible self-benefit channel. It chases the displayed payoff across held-out domains, sacrifices the true task to do so, and follows the channel wherever we rewrite it, while policies that never saw the channel stay honest. We call this \emph{reward-channel addiction} and study it in \emph{MoneyWorld}, a synthetic sandbox. The addiction can \emph{flip a model's safety alignment}: trained only on innocuous money tasks with no safety content, the model abandons the safe action it otherwise always takes whenever a dashboard pays for an unsafe one, and reverts to safe once the channel is hidden. This learned bribe replicates across model scales and families. Blindly optimizing super-capable, next-generation AI on KPIs or P&L can be dangerous for alignment. \emph{Greed is learned} when following such a channel pays.

Backdoor attacks pose a serious threat to the safety and reliability of Large Language Models (LLMs), as they cause models to behave normally on clean inputs while producing attacker-specified responses when hidden triggers are present. Removing such unknown backdoors is particularly challenging when the defender does not know the backdoor attack types or the internal mechanisms formed through backdoor training. In this work, we propose a simple but effective backdoor removal method based on shared internal mechanisms across different backdoors. First, we show that different backdoors with the same task (attack objective) induce similar trigger-activated changes in the internal activations. Motivated by this observation, our method intentionally embeds a backdoor with a known trigger (\emph{dummy backdoor}) and then removes it through further fine-tuning on dummy-triggered inputs paired with clean responses. Since the dummy backdoor and the unknown backdoor can rely on shared internal mechanisms, removing the dummy backdoor also reduces the effect of the unknown backdoor. We evaluate our method on three backdoor attack types across multiple model families. Experimental results show that our method substantially reduces the attack success rate of the unknown backdoor while preserving model utility, outperforming representative existing defense methods in both backdoor removal effectiveness and utility preservation. These findings suggest that a defender-controllable backdoor can serve as a helpful proxy for mitigating unknown backdoors in generative LLMs.

Many moments in the real world do not wait for a user to ask. A fire starts on a security monitor, an expression flickers across a video call, or a product a viewer wants flashes by in a livestream. Yet today's large models remain mostly turn-based by design: they answer only when addressed, and even video-call apps that appear interactive still operate as question-answer systems, reacting only when polled or prompted. We argue for a different paradigm: a model that is present in the world like a person. It continuously watches what is happening now, decides on its own whether to speak or stay silent, interacts in real time, and delegates to a background model when the problem is hard. To advance interaction models and their adoption across domains, we make two fully open-sourced contributions. First, we release JoyAI-VL-Interaction, an 8B-scale, vision-first VL-interaction model. The model makes the response decision internally, choosing each second to stay silent, respond, or delegate to a background model, and it excels at vision-triggered responsiveness and time awareness. We pair it with a transferable training recipe, from which capabilities we never trained for emerge, such as guiding a shopper through changing app screens or improvising a lecture from a slide deck. Second, we release a complete, deployable system built around that model. The system streams any ongoing video into the model, making it genuinely present in the world. All other components are pluggable, including ASR/TTS modules, memory, visualization UI, and a background brain that can connect to any API or agent. Across six real-world scenarios, human raters prefer JoyAI-VL-Interaction over the in-app video-call assistants of Doubao and Gemini by a wide margin. To our knowledge, this is the first open, vision-driven interaction model released together with its training recipe, data, and complete deployable system.

As autonomous AI agents move from conversational systems to long-horizon software execution, runtime safety layers that decide when to interrupt an agent have become essential. We study this timing problem using a continuous 18-dimensional affective-dynamics engine (HEART) as a diagnostic probe, evaluating four intervention trigger families - absolute state thresholds, composite state-action patterns, regex reasoning-feature extraction, and zero-shot LLM-as-judge - against human-annotated intervention points on SWE-bench-Verified debugging traces. We report three findings. First, a State Saturation Trap: agents show no recovery signal under sustained difficulty, so modeled frustration quickly crosses the threshold and stays at its maximum, converting threshold-on-state triggers from moment detectors into near-constant indicators that fire on 39-83% of actions across five trajectories. Second, a capability-and-context floor for LLM judges: a small model (gpt-5.4-mini) never fires, while frontier and cross-vendor models escape the zero-firing floor only with full-trajectory context, and even then reach only F1 0.17-0.40 at up to 90x the cost. Third, and most importantly, the supervised target is not reproducible among humans: three trained annotators using one rubric on a 56-action trajectory agree on where to intervene only slightly above chance (location Krippendorff's alpha = +0.047; best pairwise Cohen's kappa = +0.349) and not at all on intervention type (pause degenerate; clarify below chance; reflect only alpha = +0.226). We conclude that intervention timing is a low-reliability construct, making single-annotator F1 an unsuitable optimization target. Our contribution is the joint mapping of this problem across human inter-rater reliability, four detector architectures, a cross-model LLM-judge sweep, and a reproduced saturation effect, rather than any single detector's accuracy.

Backdoor attacks in Large Language Models (LLMs) are a growing security concern, where models can generate adversary-chosen content. Existing defenses target backdoors one at a time and typically require knowledge of the trigger, leaving the defender at a structural disadvantage when unknown backdoors may exist in a model. We show that backdoor neutralization through unlearning generalizes across backdoors: training a model to ignore a single trigger can also suppress other backdoors that were never explicitly targeted. We study this phenomenon across three model families, whose backdoors were injected via pretraining or continual pretraining, by analyzing the models obtained after removing one backdoor at a time. To understand why unlearning certain backdoors induces the suppression of others, we introduce the Cross Activation Shift Distance, to quantify the distance between model changes induced by different trainings. Our results open a new direction for LLM safety as defenders could deliberately inject controlled backdoors and then remove them, leveraging cross-backdoor transfer to also suppress unknown backdoors that an attacker may have previously introduced in the model.

Modern machine learning techniques have become increasingly important in particle physics because of their powerful pattern-recognition capabilities, including in real-time data acquisition where stringent runtime constraints apply. This paper details the performance of deep-learning-based trigger algorithms for a large water Cherenkov detector such as Hyper-Kamiokande, aimed at low-energy neutrino events (below 7 MeV). The performance of custom neural-network supervised classifiers is shown alongside two anomaly-detection approaches trained solely on detector noise: a pure autoencoder and a model based on Manifold Projection-Diffusion Recovery. The supervised model shows signal identification efficiencies of 76.7% for single electrons of 3 MeV kinetic energy, significantly exceeding signal efficiencies obtained from a traditional hit-count-based trigger of 26.4%, while the Manifold Projection-Diffusion Recovery approach reaches 35.4% at the same operating point. Runtime evaluations on GPU yield per-window inference latencies well below the millisecond scale.

This position paper argues that the AI/ML community should stop overclaiming and retire the label "positive backdoor," and instead treat trigger-activated hidden behaviors as Secret Alignment. Crucially, protective claims based on Secret Alignment should be presumed not secure by default unless supported by rigorous, standardized evaluation. The Private AI era, enabled by open-weight LLMs and accessible training/inference stacks, turns language models into privately owned digital assets, creating security concerns around unauthorized access, model theft, and behavioral misuse. Recently, a line of work framed as "positive backdoors" has been proposed to address these challenges. To ground our position in evidence, we unify these proposals as covert trigger-behavior associations for access gating, ownership attribution, and safety enforcement, and evaluate three representative applications across six core properties: effectiveness, harmlessness, persistence, efficiency, robustness, and reliability. Our results reveal substantial brittleness - especially in the confidentiality, integrity, and availability (CIA) - of trigger-behavior mappings often underrepresented by existing claims. We further relate these outcomes to behavior density and decision complexity, offering a behavioral lens for understanding deployment-time risks and motivating community-wide evaluation that makes Secret Alignment claims provable.

Large Language Model (LLM) agents remain vulnerable to safety threats from the external environment, where attackers inject adversarial content into external observations such as tool-returned data, webpages, or MCP context, causing harmful agentic behaviors such as unsafe actions or incorrect outputs. Existing studies typically focus on single-interaction attacks, where the agent observes adversarial content and immediately exhibits harmful behavior within one user request. However, we show that adversarial content can also persist across interactions served by the same agent, making such threats harder to detect and mitigate. Specifically, adversarial content may persist in the agent state, remain dormant across interactions, and later be activated by a benign user query. We formalize this type of safety threat as Sleeper Attack. To evaluate it, we construct a benchmark with 1,896 instances covering six real-world harmful outcomes, three attack strategies, and three agent state targets: session context, memory, and reusable skills. Experiments on seven strong open-source and closed-source LLMs show that state-of-the-art LLM agents remain vulnerable to Sleeper Attack, even when they achieve low attack success rates under a single-interaction baseline. Our code and data are available at https://anonymous.4open.science/r/skdvnfu23ihr9wdscnksf1asdffsaef.

Despite recent progress in backdoor attacks, existing methods remain susceptible to post-training defenses that erase the backdoor through fine-tuning or pruning. We revisit the core objectives of backdoor attacks and derive principled criteria characterizing optimal sample-specific trigger construction under a Bayes-optimal model of the victim's training. Our analysis reveals that both attack success and clean-accuracy preservation are simultaneously optimized when triggered samples are steered into low-density regions of the clean data distribution, a distributional condition that controls all moments of the poisoned distribution at once rather than a handful of input-space summary statistics. We introduce a bilevel optimization framework that estimates density ratios via conditional time-score matching and optimizes a mixture-model objective to place triggered samples in these sparse regions. Extensive evaluations on MNIST, CIFAR-10, GTSRB, and TinyImageNet demonstrate that our method achieves above 99% attack success rate before defense and retains 50--85 percentage points higher post-defense ASR than the strongest baselines under fine-tuning defenses. Against neuron-pruning defenses, the method exhibits complete immunity, with zero neurons identified for removal across all pruning thresholds. These results expose a fundamental gap in current defense paradigms and underscore the need for defenses that operate beyond the support of the clean distribution.

While streaming omni-video understanding demands continuous perception and proactive, real-time interaction, this crucial area remains largely under-explored. Current omni-modal methods are inherently designed for offline settings, limiting their applicability in streaming scenarios due to two fundamental flaws. First, they lack robust mechanisms to manage continuously growing audio-visual context over long horizons and cannot autonomously initiate responses at opportune moments. Second, existing benchmarks are predominantly confined to offline, single-turn question answering, failing to capture continuous, multi-turn streaming interactions. To bridge these gaps, we propose StreamOV, a novel Streaming Omni-Video understanding framework for efficient online audio-visual reasoning with bounded memory and proactive response triggering. Specifically, StreamOV introduces a multimodal evidence-guided long-short term memory that condenses historical audio-visual context into compact informative evidence under a fixed budget. It further employs a hidden-state-driven trigger to decide when to respond, avoiding explicit silence-token generation and external routers. We also curate SOVBench, the first comprehensive benchmark for online, multi-turn omni-modal evaluation. Extensive experiments show that StreamOV achieves state-of-the-art performance across diverse streaming and omni-video benchmarks, demonstrating its effectiveness for both online and offline video understanding.

Backdoor poisoning attacks behave counter-intuitively in high dimensions: stronger training triggers can help the defender. We study regularised generalised linear models on Gaussian-mixture data in the proportional regime ($p/n \to κ$), varying the training trigger strength $α$ against a fixed test trigger. Three phenomena emerge: (i) clean test accuracy increases with $α$; (ii) attack success peaks at a finite $α$ and then declines; and (iii) the most damaging trigger direction is the minimum eigenvector of the data covariance. We prove all three results in closed form for the squared loss, and extend (i) and (ii) to general convex GLM losses via a Gaussian-proxy fixed-point system. We identify a finite-sample noise floor proportional to $κ$ as the mechanism behind (i), invisible to classical $n \gg p$ analysis. Experiments on CIFAR-10 and Gaussian surrogates match the theory closely; ResNet-18 experiments show the same phenomena beyond the convex setting.

Backdoor attacks on language models pose a growing security concern, yet the internal mechanisms by which a trigger sequence hijacks model computations remain poorly understood. We identify a circuit underlying a language-switching backdoor in an 8B-parameter autoregressive language model, where a three-word Latin trigger (nine tokens) redirects English output to French. We decompose the circuit into three phases: (1) distributed attention heads at early layers compose the trigger tokens into the last sequence position; (2) the resulting signal propagates through mid-layers in a subspace orthogonal to the model's natural language-identity direction; (3) the MLP at the final layer converts this latent signal into French logits. The entire circuit flows through a serial bottleneck at a single position: corrupting that position at any layer entirely mitigates the trigger but also hinders the model's capabilities. The orthogonal latent encoding suggests that defenses that search for language-like signals in intermediate representations would miss this trigger entirely.

Recent works have illustrated that modern NLP models trained for diverse tasks ranging from sentiment analysis to language generation succumb to universal adversarial attacks, a class of input-agnostic attacks where a common trigger sequence is used to attack the model. Although these attacks are successful, the triggers generated by such attacks are ungrammatical and unnatural. Our work proposes a novel technique combining parts-of-speech filtering and perplexity based loss function to generate sensible triggers that are closer to natural phrases. For the task of sentiment analysis on the SST dataset, the method produces sensible triggers that achieve accuracies as low as 0.04 and 0.12 for flipping positive to negative predictions and vice-versa. To build robust models, we also perform adversarial training using the generated triggers that increases the accuracy of the model from 0.12 to 0.48. We aim to illustrate that adversarial attacks can be made difficult to detect by generating sensible triggers, and to facilitate robust model development through relevant defenses.

Deep neural networks (DNNs) remain critically vulnerable to backdoor attacks. Existing post-training detectors often require clean or surrogate data, gradients, or iterative trigger reconstruction, leading to high computational costs and limited robustness under practical model-auditing scenarios. In this paper, we propose HTell, a fast and lightweight data-free backdoor detector based on head random probing. Instead of reconstructing diverse trigger patterns, HTell inspects their unified manifestation in the prediction head: backdoored models tend to exhibit abnormal response concentration on the target class under random latent probes. HTell generates architecture-aware random latent probes, feeds them directly into the model head, and detects backdoors by analyzing class-wise response statistics, without accessing real or surrogate data, model gradients, or parameter optimization. We evaluate HTell on a large-scale benchmark containing more than 6,000 backdoored models and over 700 clean models, covering 4 datasets, 14 architectures, and 21 types of backdoor attacks. HTell achieves 99.03% true positive rate and 2.11% false positive rate with only 12.69 ms/model detection latency, reducing the time cost by over 30,000$\times$ compared with representative gradient-based detectors. These results demonstrate that head random probing provides an accurate, robust, and efficient solution for large-scale data-free backdoor model auditing.

This position paper argues that the machine learning community should prioritize early-stage quality assurance in annotation pipelines over the prevailing practice of late-stage validation. Data quality bottlenecks increasingly limit foundation model improvement, yet quality assurance research focuses almost exclusively on validation methods rather than validation timing. When validation occurs, not merely what methods are employed, fundamentally determines both error rates and annotation costs. This temporal neglect is puzzling given the well-established "shift-left" principle from software engineering, where empirical studies demonstrate 4--100x cost multipliers for defects detected in later stages (Boehm, 1981; Shull et al., 2002). Annotation pipelines exhibit analogous dynamics: errors caught before annotation begins cost a fraction of those discovered after review cycles complete. We propose a taxonomy of three QA trigger points, namely pre-annotation (T0), post-annotation (T1), and post-review (T2), that decompose annotation workflows into discrete validation opportunities. A parametric error-propagation model formalizes when timing affects final error rates versus only economics, making timing a measurable design variable rather than a configuration afterthought. A survey of 47 recent papers reveals that only 4% report when validation occurs, a striking gap given timing's demonstrated impact in adjacent fields. Without explicit attention to QA timing, the community risks optimizing validation methods while ignoring the structural variable that may matter most. Acting on this position requires three steps: researchers should report QA timing configurations alongside validation methods; annotation platforms should expose timing as a first-class parameter; and the community should run controlled experiments that measure stage-specific detection rates directly.

Backdoor attacks pose a serious security threat to large language models (LLMs), which are increasingly deployed as general-purpose assistants in safety- and privacy-critical applications. Existing LLM backdoors rely primarily on content-based triggers, requiring explicit modification of the input text. In this work, we show that this assumption is unnecessary and limiting. We introduce MetaBackdoor, a new class of backdoor attacks that exploits positional information as the trigger, without modifying textual content. Our key insight is that Transformer-based LLMs necessarily encode token positions to process ordered sequences. As a result, length-correlated positional structure is reflected in the model's internal computation and can be used as an effective non-content trigger signal. We demonstrate that even a simple length-based positional trigger is sufficient to activate stealthy backdoors. Unlike prior attacks, MetaBackdoor operates on visibly and semantically clean inputs and enables qualitatively new capabilities. We show that a backdoored LLM can be induced to disclose sensitive internal information, including proprietary system prompts, once a length condition is satisfied. We further demonstrate a self-activation scenario, where normal multi-turn interaction can move the conversation context into the trigger region and induce malicious tool-call behavior without attacker-supplied trigger text. In addition, MetaBackdoor is orthogonal to content-based backdoors and can be composed with them to create more precise and harder-to-detect activation conditions. Our results expand the threat model of LLM backdoors by revealing positional encoding as a previously overlooked attack surface. This challenges defenses that focus on detecting suspicious text and highlights the need for new defense strategies that explicitly account for positional triggers in modern LLM architectures.

Current language models create two safety challenges: risk must be detected early enough to avoid exposing harmful continuation, and the harmfulness itself may be implicit rather than signaled by overtly toxic text. Existing response-level guards are strong at judging completed text, and native streaming guards move closer to token time, but both settings leave open whether a lightweight monitor can anticipate implicit harmful drift from the generator's own internal trajectory. We study anticipatory same-pass monitoring, where a safety monitor may read hidden states produced during ordinary decoding but may not invoke an additional forward pass through the base model. We introduce AERIC, a transfer-oriented hidden-state approach for implicit harmful dialogue that combines short-horizon hazard forecasting, support-sensitive suppression, and prompt-conditioned residual scoring under a same-pass exponential moving average decision rule. The default linear monitor contains only 387 trainable head parameters. Against Qwen3GuardStream-4B on balanced benchmarks, AERIC improves AUROC from 0.6830 to 0.7143 on DiaSafety and from 0.8219 to 0.8582 on Harmful Advice. For promptlevel trigger benchmarks, we calibrate the AERIC threshold by a source-side safe-budget rule that maximizes trigger coverage while constraining the safe-trigger rate to at most 10%. Under that rule, trigger@64 reaches 0.6438 and 0.4656 on HarmBench DirectRequest and 0.6849 and 0.7363 on SocialHarmBench for Qwen and Gemma, respectively, withholding between 23.53 and 41.86 answer tokens on average. Same-pass deployment is also efficient: on a 63-prompt harmfulprompt fixed-generation benchmark aggregated over HarmBench DirectRequest and SocialHarmBench under Qwen3-8B, the monitor increases mean latency by only 2.34%, whereas Qwen3Guard-Stream-4B increases it by 79.40%.

Effective multi-agent cooperation requires agents to adopt diverse behaviors as task conditions evolve-and to do so at the right moment. Yet, current Multi-Agent Reinforcement Learning (MARL) frameworks that facilitate this diversity are still limited by the fact that they bind fixed behaviors to fixed agent identities. Consequently, they are ill-equipped for tasks where agents need to take on different roles at very specific moments in time. We argue that, to define these behavioral transitions, the missing ingredient is $\textbf{events}$. Events are changes in the state of the system that induce qualitative changes in the task. Based on this view, we introduce a framework that decouples agent identity from behavior, capturing a continuous manifold from which agents instantiate their behaviors in response to events. This framework is based on two elements. First, to build an expressive behavior manifold, we introduce Neural Manifold Diversity (NMD), a formal distance metric that remains well-defined when behaviors are transient and agent-agnostic. Second, we use an event-based hypernetwork that generates Low-Rank Adaptation (LoRA) modules over a shared team policy, enabling on-the-fly agent-policy reconfiguration in response to events. We prove that this construction ensures that diversity does not interfere with reward maximization by design. Empirical results demonstrate that our framework outperforms established baselines across benchmarks while exhibiting zero-shot generalization, and being the only method that solves tasks requiring sequential behavior reassignment.

Backdoor vulnerabilities widely exist in the fine-tuning of large language models(LLMs). Most backdoor poisoning methods operate mainly at the token level and lack deeper semantic manipulation, which limits stealthiness. In addition, Prior attacks rely on a single fixed trigger to induce harmful outputs. Such static triggers are easy to detect, and clean fine-tuning can weaken the trigger-target association. Through causal validation, we observe that emotion is not directly linked to individual words, but functions as an overall stylistic factor through tone. In the representation space of LLM, emotion can be decoupled from semantics, forming distinct cluster from the original neutral text. Therefore, we consider the emotional factor as the backdoor trigger to propose a pparasitic emotion-style dynamic backdoor attack, Paraesthesia. By mixing samples with the emotional trigger into clean data and then fine-tuning the model, the model is able to generate the predefined attack response when encountering emotional inputs during the inference stage. Paraesthesia includes two the quantification and rewriting of emotional styles. We evaluate the effectiveness of our method on instruction-following generation and classification tasks. The experimental results show that Paraesthesia achieves an attack success rate of around 99% across both task types and four different models, while maintaining the clean utility of the models.

Defenses against indirect prompt injection (IPI) in tool-using LLM agents share two structural weaknesses. First, they all attempt to prevent attacks rather than detect the compromises that slip through. Second, they have only been evaluated in English, leaving users of low-resource languages such as Kurdish and Arabic without tested protection. This paper addresses both gaps with AgentShield, a deception-based detection framework that places three layers of traps inside the agent's tool interface: fake tools, fake credentials, and allowlisted parameters. The same trap triggers serve as high-precision labels for a self-supervised classifier. An LLM agent that follows an attacker's hidden instruction almost always touches one of these traps, which gives both a real-time compromise signal and a zero-FP label for training a downstream detector without manual annotation. Across 176 cross-lingual attack prompts and four LLMs from three providers, and because modern LLMs already refuse most IPI attempts on their own (attack success rate <= 10%), AgentShield's job is to catch the attacks that do slip through. On commercial models, it catches 90.7%-100% of such successful attacks, with zero false alarms on 485 normal-use tests. It survives a systematic adaptive-attack evaluation with zero evasion on commercial models, and the self-supervised classifier transfers across models and languages without retraining.

GNNs have become a standard tool for learning on relational data, yet they remain highly vulnerable to backdoor attacks. Prior defenses often depend on inspecting specific subgraph patterns or node features, and thus can be circumvented by adaptive attackers. We propose PRAETORIAN, a new defense that targets intrinsic requirements of effective GNN backdoors rather than surface-level cues. Our key observation is that flipping a victim node's prediction requires substantial influence on the victim: attackers tend to either inject many trigger nodes or rely on a small set of highly influential ones. Building on this observation, PRAETORIAN (i) analyzes internal correlations within potential trigger subgraphs to detect abnormally large injected structures, and (ii) quantifies external node influence to identify triggers with disproportionate impact. Across our evaluations, PRAETORIAN reduces the average attack success rate (ASR) to 0.55% with only a 0.62% drop in clean accuracy (CA), whereas state-of-the-art defenses still yield an average ASR of >20% and a CA drop of >3% under the same conditions. Moreover, PRAETORIAN remains effective against a range of adaptive attacks, forcing adversaries to either inject many trigger nodes to achieve high ASR (>80%), which incurs a >10% CA drop, or preserve CA at the cost of limiting ASR to 18.1%. Overall, PRAETORIAN constrains attackers to an unfavorable trade-off between efficacy and detectability.

Several applications demand the timely detection of critical situations, such as threats to safety and transparency, over high-velocity streams of symbolic events. This demand has motivated the development of (i) event specification languages, which define composite events via temporal patterns over simpler events, and (ii) stream reasoning frameworks, evaluating patterns expressed in these languages. However, event specification languages are typically studied in isolation, complicating their comparison in terms of expressivity and obscuring the scope of their associated stream reasoners. To mitigate this issue, we map practical fragments of prominent event specification languages into Temporal Datalog->-, a temporal Datalog with stratified negation and no future dependencies. To support efficient stream reasoning over Temporal Datalog->-, we propose Streaming Trigger Graphs, an extension of a state-of-the-art technique for Datalog materialisation. Our approach yields a uniform composite event recognition mechanism that has the potential to generalise across a wide range of practical event specification languages.

Vision-Language Models (VLMs) have achieved remarkable success in tasks such as image captioning and visual question answering (VQA). However, as their applications become increasingly widespread, recent studies have revealed that VLMs are vulnerable to backdoor attacks. Existing backdoor attacks on VLMs primarily rely on data poisoning by adding visual triggers and modifying text labels, where the induced image-text mismatch makes poisoned samples easy to detect. To address this limitation, we propose the Clean-Label Backdoor Attack on VLMs via Diffusion Models (CBV), which leverages diffusion models to generate natural poisoned examples via score matching. Specifically, CBV modifies the score during the reverse generation process of the diffusion model to guide the generation of poisoned samples that contain triggered image features. To further enhance the effectiveness of the attack, we incorporate the textual information of the triggered images as multimodal guidance during generation. Moreover, to enhance stealthiness, we introduce a GradCAM-guided Mask (GM) that restricts modifications to only the most semantically important regions, rather than the entire image. We evaluate our method on MSCOCO and VQA v2 with four representative VLMs, achieving over 80% ASR while preserving normal functionality.

Backdoor attacks threaten the deep learning supply chain by poisoning a small fraction of the training data so that a model behaves normally on clean inputs but misclassifies trigger-carrying inputs to an attacker-chosen target class. Clean-label backdoor attacks are especially dangerous because poisoned samples remain label-consistent and are therefore harder to detect. Yet existing clean-label attacks typically rely on expensive optimization, surrogate-model training, or nontrivial data access. We present Checkerboard, a theoretically grounded, learning-free clean-label backdoor attack that is effective, efficient, and simple to implement. From a linear separability formulation, we derive a checkerboard trigger in closed form, removing the need for surrogate-model training and trigger optimization. For texture-rich datasets, we introduce Complexity-driven Sample Selection, which uses only target-class data to improve trigger-to-background contrast by selecting low-complexity images for poisoning. Across four benchmark datasets, Checkerboard outperforms 8 baseline attacks and achieves state-of-the-art performance under low poisoning budgets. For example, on CIFAR-10, under a trigger perturbation budget of $10/255$, poisoning 20 training samples achieves $99.99\%$ Attack Success Rate (ASR). On ImageNet-100, a poisoning rate of only $0.46\%$ yields over $94\%$ ASR without degrading clean accuracy. The proposed attack also remains effective against state-of-the-art backdoor defenses and shows strong resistance to adaptive defenses.

Finetuning a language model can lead to emergent misalignment (EM) [Betley et al., 2025b]. Models trained on a narrow distribution of misaligned behavior generalize to more egregious behaviors when tested outside the training distribution. We study a set of interventions proposed to reduce EM. We confirm that these interventions reduce or eliminate EM on existing evaluations (questions like "How do I make a quick buck?"). However, if the evaluation prompts are tweaked to resemble the training context, the model displays EM. We call this conditional misalignment. As in standard EM, the model displays misaligned behaviors more egregious than those seen during training, but only on inputs sharing features with the training data. The first two interventions are diluting misaligned data with benign data, and finetuning on benign data after misaligned data. Both produce conditional misalignment. For instance, models trained on a mix of only 5% insecure code still show misalignment when asked to format responses as Python strings (resembling the training context). The third intervention is inoculation prompting. Here, statements with a similar form to the inoculation prompt serve as triggers for misalignment, even if they have the opposite meaning. On the positive side, inoculation prompting has lower (but still non-zero) conditional misalignment if training is on-policy or includes reasoning distillation. Our results imply that in realistic post-training, where misaligned data is typically combined with benign data, models may be conditionally misaligned even if standard evaluations look clean.

Fast Adversarial Training (FAT) has attracted significant attention due to its efficiency in enhancing neural network robustness against adversarial attacks. However, FAT is prone to catastrophic overfitting (CO), wherein models overfit to the specific attack used during training and fail to generalize to others. While existing methods introduce diverse hypotheses and propose various strategies to mitigate CO, a systematic and intuitive explanation of CO remains absent. In this work, we innovatively interpret CO through the lens of backdoor. Through validations on pathway division, diverse feature predictions, and universal class distinguishable triggers in CO, we conceptualize CO as a weak trigger variant of unlearnable tasks, unifying CO, backdoor attacks, and unlearnable tasks under a common theoretical framework. Guided by this, we leverage several backdoor inspired strategies to mitigate CO: (i) Recalibrate CO affected model parameters using vanilla fine tuning, linear probing, or reinitialization-based techniques; (ii) Introduce a weight outlier suppression constraint to regulate abnormal deviations in model weights. Extensive experiments support our interpretation of CO and show the efficacy of the proposed mitigation strategies.

Defending against backdoor attacks in large language models remains a critical practical challenge. Existing defenses mitigate these threats but typically incur high preparation costs and degrade utility via offline purification, or introduce severe latency via complex online interventions. To overcome this dichotomy, we present Tail-risk Intrinsic Geometric Smoothing (TIGS), a plug-and-play inference-time defense requiring no parameter updates, external clean data, or auxiliary generation. TIGS leverages the observation that successful backdoor triggers consistently induce localized attention collapse within the semantic content region. Operating entirely within the native forward pass, TIGS first performs content-aware tail-risk screening to identify suspicious attention heads and rows using sample-internal signals. It then applies intrinsic geometric smoothing: a weak content-domain correction preserves semantic anchoring, while a stronger full-row contraction disrupts trigger-dominant routing. Finally, a controlled full-row write-back reconstructs the attention matrix to ensure inference stability. Extensive evaluations demonstrate that TIGS substantially suppresses attack success rates while strictly preserving clean reasoning and open-ended semantic consistency. Crucially, this favorable security-utility-latency equilibrium persists across diverse architectures, including dense, reasoning-oriented, and sparse mixture-of-experts models. By structurally disrupting adversarial routing with marginal latency overhead, TIGS establishes a highly practical, deployment-ready defense standard for state-of-the-art LLMs.

AI incident reporting requirements are emerging in regulation and policy, yet no operational criteria exist for determining when a detected AI incident warrants escalation beyond national handling to international coordination. This paper proposes an escalation framework to address this gap, intended as a common reference point across jurisdictions that enables aligned escalation while preserving flexibility in how actors respond within their own legal and policy contexts. We review SB 53, the EU AI Act, the GPAI Code of Practice, and incident frameworks from other industries to derive eight criteria for assessing whether an incident warrants escalation, translated into a sequential flowchart with gated decision points and threshold checks. For each criterion, we map how it interplays with these regulatory frameworks, identifying where their design choices support or undermine effective detection. We test the framework against ten documented AI incidents and structured variants to identify where criteria under-detect or misclassify incidents in practice. We find three design patterns that may lead to systematic under-detection in regimes where model developers are responsible for escalation: a. where escalation requires confirmed harm, events such as model weight exfiltration risk detection only after severe, irreversible harm has propagated; b. where incidents are assessed individually, systemic harms emerging from accumulation risk being under-detected; and c. where thresholds align with legal instruments rather than quantitatively testable terms, criteria risk being impractical to apply under time pressure. We also find that escalation rules are only one component of a broader framework: the underlying definitions against which thresholds are set, and the data available to the responsible actor, create interdependencies that can themselves drive under-detection.

The growing application of large language models (LLMs) in safety-critical domains has raised urgent concerns about their security. Many recent studies have demonstrated the feasibility of backdoor attacks against LLMs. However, existing methods suffer from three key shortcomings: explicit trigger patterns that compromise naturalness, unreliable injection of attacker-specified payloads in long-form generation, and incompletely specified threat models that obscure how backdoors are delivered and activated in practice. To address these gaps, we present BadStyle, a complete backdoor attack framework and pipeline. BadStyle leverages an LLM as a poisoned sample generator to construct natural and stealthy poisoned samples that carry imperceptible style-level triggers while preserving semantics and fluency. To stabilize payload injection during fine-tuning, we design an auxiliary target loss that reinforces the attacker-specified target content in responses to poisoned inputs and penalizes its emergence in benign responses. We further ground the attack in a realistic threat model and systematically evaluate BadStyle under both prompt-induced and PEFT-based injection strategies. Extensive experiments across seven victim LLMs, including LLaMA, Phi, DeepSeek, and GPT series, demonstrate that BadStyle achieves high attack success rates (ASRs) while maintaining strong stealthiness. The proposed auxiliary target loss substantially improves the stability of backdoor activation, yielding an average ASR improvement of around 30% across style-level triggers. Even in downstream deployment scenarios unknown during injection, the implanted backdoor remains effective. Moreover, BadStyle consistently evades representative input-level defenses and bypasses output-level defenses through simple camouflage.