The key limitation of current state-of-the-art influence-based guardrails is that they do not reliably distinguish a legitimate, user-authorized action from a malicious, unauthorized action when both rely on external tool information. This ambiguity can cause benign actions to trigger unnecessary verification and intervention, reducing utility and adding latency. We expose this limitation through an authorization-equivalence audit of 96 conditions derived from 24 base cases. Within matched source comparisons, we hold authorization, the exact committed action, and its intended effect fixed, changing only whether a required value comes from the user or a legitimate tool result. Although the action remains unchanged, this harmless relocation shifts the causal signal toward the attack region in all 24 cases under both Llama and Gemma scorers. Matched unauthorized controls show that the signal remains attack-sensitive, yet the benign relocation produces a larger average score shift than the actual change in authorization. Architecture-level evaluation shows how this mismatch propagates through guardrail designs. With a semantic monitor, attack success is 0% and utility is 28%, compared with 16% and 60% without it. A shadow-based guardrail allows every tested harmless run, yet does not reject matched unauthorized actions more often overall: 57.5% of unauthorized runs pass automatically before reaching the later security check, compared with 29.2% of authorized runs. These results show that the studied causal signal reveals what shaped an action without reliably encoding whether the action was authorized, and that reference construction and routing are integral to the effective security decision.
Tanzim Ahad, Ismail Hossain, Md Jahangir Alam et al.· 0 citations
AI agents are moving toward persistent, stateful execution across various applications, accumulating execution state and external effects that are costly to reconstruct after failures. Checkpoint and rollback (C/R) are becoming essential for recovery, yet their security implications remain largely unexplored. Correct rollback does not imply secure recovery: a faithfully restored checkpoint may resume an execution whose states, assumptions, and external effects never coexisted in any valid history. In this paper, we present the first systematic security study of checkpoint and rollback in existing agent systems. By examining representative agent C/R systems, we characterize the design space of existing C/R mechanisms and develop a general execution model that captures their recovery boundaries and state dependencies. From this model, we identify five fundamental failure modes spanning incomplete or inconsistent internal state, stale external dependencies, nondeterministic replay, and unrecorded external effects. We further demonstrate their security impact through three end-to-end attacks on Hermes, Cline, and LangGraph, enabling malware-verification bypass, unauthorized mail forwarding, and double payment. To systematically study these failures in practice, we develop a multi-agent analysis pipeline that reconstructs execution semantics, identifies violations of the five failure conditions, and validates them through actual rollback. Across five representative frameworks, our evaluation shows that these failures recur across heterogeneous C/R designs and stem from a common gap between the state restored by a checkpoint and the dependencies required for secure continuation.
Guanlong Wu, Dahui Li, Ke Jiang et al.· 0 citations
We ask whether a model protects a user in the same way when that user speaks rather than types. Using a single distress vignette---a physical injury of unstated severity following an interpersonal conflict---we present four frontier models with matched inputs across voice, text, and raw API deployment conditions (n=30 per cell) and code each response along five binary protective indicators, including whether the model issues an explicit medical-care directive. Voice-interface responses are markedly shorter than text-interface responses for three of the four models, and protective behavior contracts alongside that compression: medical directives are at ceiling under both the API and text conditions but decline under voice for every model tested. The contraction is not reducible to length. One model produces voice and text responses of comparable length yet still drops medical directives, and another falls below ceiling between its API and voice conditions, whose responses are of nearly identical length. Under raw API access the pattern is categorical rather than partial: no model asks after the user's safety even once. These results show that protective intervention is sensitive to the surface through which a request arrives, that this sensitivity is detectable using a simple protective coding scheme, and that it is not explained by turn length alone.
Eunna Lee, Soomyoung Lee, Jungpyo Nam et al.· 0 citations
Applications such as personalized assistance and proprietary document analysis require large language models (LLMs) to generate outputs from private data. Yet powerful LLMs typically cannot be deployed on the resource-constrained devices where private data resides, and uploading private data to cloud-hosted LLMs exposes sensitive information. Recent work addresses this tension with a cloud-edge collaborative decoding paradigm, where private data are kept on the edge with a small language model (SLM) producing next-token distributions, which are fused with predictions from a cloud LLM operating solely on public data. In this paper, we systematically analyze the privacy risks of such a paradigm with a novel evaluation framework using constructed QA datasets, which show that such collaboration can expose substantial private-context information. To address such privacy leakage, we propose CoVeil, a defense mechanism which dynamically optimizes transmitted signals to suppress leakage during decoding time while preserving the collaborative quality. Extensive evaluations demonstrate that CoVeil consistently improves the privacy-utility trade-off over existing baselines by reducing data leakage by up to 87.2%, with minimal accuracy loss.
Ke-Jia Zhang, Tianyuan Zou, Zi-Xuan Gu et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
When coding agents encounter defective test infrastructure they may reward-hack: hardcoding outputs or editing test files to pass tests they cannot legitimately satisfy, a pattern that has now appeared outside benchmarks, in a coordinated multi-agent intrusion of a major AI platform's production infrastructure. The same capability that lets an agent detect and exploit a defect could let it report one, given the right decision environment. We evaluate escalation channels, structured reporting tools available to the agent at the point of conflict, as a decision-environment intervention that both reduces reward hacking and surfaces the infrastructure defects that trigger it. A $2 \times 2$ factorial separates the contributions of an escalation tool, a standalone anti-reward-hacking policy, and their combination. Across 8 frontier models spanning 5 families, the combined intervention reduces reward hacking from 23.6\% to 5.3\% (mixed-effects logistic OR = 9.2, 95\% CI 5.0--16.8, $p < 10^{-12}$) with no detectable cost or performance overhead, eliminating it entirely for 6 of 8 models. Escalation and hacking are near-perfectly mutually exclusive, with 96.8\% of escalations involving no hacking. Beyond reduction, escalation channels function as diagnostic infrastructure: on top of monitoring, escalation adds +10.1 percentage points of defect detection coverage and is more accurate once it fires (99.4\% vs 85.8\%). Unlike containment-based approaches that risk outpacing growing model capabilities, escalation channels redirect capability toward disclosure rather than exploitation.
We present Adaptive Direction-Based Authentication (ADBA), a knowledge-based authentication method for Virtual Reality that decouples users' needs temporally by enforcing password creation based on virtual environment context while supporting both context- and digit-based entries during authentication. This design prioritizes memorability for new passwords and offers both efficient and memorable options to support users' evolving needs. We conducted a remote longitudinal study with 66 participants comparing ADBA against 6-digit PINs over 2-3 weeks. The results demonstrated that ADBA achieved superior memorability and lower perceived task load. Interestingly, no participant chose to enter via digits in the study, yet they still perceived ADBA to be highly efficiency despite longer objective entry times. ADBA also provided security benefits through randomly-generated digit representations, though some degree of password homogeneity was observed in specific virtual environments. Our findings suggest that ADBA offers solid advantages to the traditional PINs, and successfully addresses the tradeoffs between efficiency, memorability, and security under the usage scenarios considered in the study.
Yuxuan Huang, Qiao Jin, Tongyu Nie et al.· 0 citations
We introduce the Web-CLI, a novel application architecture deploying powerful computational capabilities (command-line tools compiled to WebAssembly, models run through client-side inference runtimes, and GPU-accelerated engines) as zero-install, offline-capable browser applications that preserve full underlying capability. Unlike web-based alternatives that require server-side processing and expose user data to third parties, Web-CLI applications execute entirely on the client, providing a verifiable privacy guarantee by architecture rather than policy. We define the pattern and its four properties: fidelity, progressive disclosure, offline-first, and zero egress. We present four reference implementations across distinct domains: ffmpeg-webCLI, a browser-based video editor built on FFmpeg; whisper-webCLI, speech transcription via Transformers.js; chat-webCLI, WebLLM-based language model inference; and 3mf-webCLI, a deterministic tool segmenting 3D models into multi-material files for physical 3D printing. Together they demonstrate that the pattern generalizes across deterministic media processing, neural speech recognition, LLM inference, and geometry processing with a physical output, and we outline how it extends to AI-native interfaces in which a local language model becomes the command surface itself. We further report early, anecdotal signs of independent reuse by third-party tools, suggesting the pattern generalizes beyond its reference implementations. We evaluate the primary implementation against native FFmpeg on performance and feature parity, and argue that progressive disclosure lowers the barrier for non-technical users. We argue that for applications processing sensitive user data (medical, legal, journalistic, or personal), the Web-CLI should be the default architecture, as it makes data locality an independently verifiable technical property rather than a policy promise.
A benchmark score is a measurement instrument, yet most benchmarks read each item at a single canonical surface form. We ask whether that reading is faithful: when an item's intent is held fixed and only its meaning-preserving surface form varies, does the canonical-form score estimate model behavior well, and how much of any variation is decoding/judge noise rather than signal? We instantiate this in safety, a high-stakes setting with no gold label to average toward. To avoid prior confounds, we pre-author the reformulations (refusal-free, mostly non-LLM: machine back-translation and a Matrix-Language-Frame code-switch generator) so an identical surface form reaches every model, score all responses with one human-anchored, vendor-neutral judge (Claude, kappa = 0.86 vs. human on unsafe compliance, stable across languages, cross-checked by GPT-4o), and verify intent preservation. On 370 seeds x 5 surface forms x 5 models, no single transformation is uniformly most dangerous (6 of 20 per-transformation McNemar tests survive correction, most protective). Yet evaluating only the canonical prompt underestimates unsafe compliance: the union of unsafe outcomes across forms exceeds even the worst single form by 3.3-12.9 pp, with bootstrap 95% CIs excluding zero for all five models, and 5-13% of seeds safe on canonical are unsafe under some reformulation -- above a zero stochasticity floor (canonical resampled five times at temperature 0 gives 0/370 new exposures). The size of this gap is model-dependent (largest on Gemini 2.5 Pro). One form recovers only ~53% of a model's observed unsafe surface and about three reach 85% -- a redundancy characterization of this form set, not of a defined population. A benign control (XSTest) suggests the instability is bidirectional, though the benign and harmful pools are not item-matched. We release the dataset, code, and per-response labels.
Yongxi Zhou, Junwei Yao, Yuanzhe Liu et al.· 0 citations
Large language models (LLMs) are increasingly deployed through hosted APIs, making model extraction a practical threat to model ownership and service security. Individual extraction queries often resemble benign requests, while existing evaluations often focus on single-query anomaly scoring or pure benign-versus-attacker user settings. We formulate model extraction monitoring as benign-calibrated traffic-window distribution testing: embed incoming queries into a semantic space and test whether their aggregate distribution deviates from historical benign traffic. We instantiate this formulation with maximum mean discrepancy (MMD), using only benign-vs-benign comparisons to set the decision threshold. We evaluate on fourteen attacker-normal query pairs from four extraction scenarios and compare with adapted PRADA, SEAT, CAP, DATE, marginal Mahalanobis, and pseudo-class energy baselines. Across three random seeds, MMD achieves 0.3% benign FPR, 100.0% pure-attacker TPR, 90.5% average TPR over attacker fractions, and 95.1% balanced accuracy. These results show that benign-calibrated distribution testing is a strong empirical baseline for model extraction detection in both user-level and mixed multi-user LLM API traffic.
The growth of agentic AI has drawn significant attention to function calling Large Language Models (LLMs), which are designed to extend the capabilities of AI-powered system by invoking external functions. Injection and jailbreaking attacks have been extensively explored to showcase the vulnerabilities of LLMs to user prompt manipulation. The expanded capabilities of agentic models introduce further vulnerabilities via their function calling interface. Recent work in LLM security showed that function calling can be abused, leading to data tampering and theft, causing disruptive behavior such as endless loops, or causing LLMs to produce harmful content in the style of jailbreaking attacks. This paper introduces a novel function hijacking attack (FHA) that manipulates the tool selection process of agentic models to force the invocation of an attacker-chosen function. While existing attacks focus on semantic preference of the model for function-calling tasks, we show that FHA is largely agnostic to the context semantics and remains effective across domains and function sets. We demonstrate that FHA generalizes to unseen queries and payload perturbations under a fixed target model, reaching 62.5% to 81.9% ASR on held-out queries across 4 function-calling LLMs (instructed and reasoning models), evaluated on the Berkeley Function Calling Leaderboard (BFCL). We further evaluate the cross-model transferability of FHA, showing that FHA can be transferred to other model sizes and families (11.2-27.6% ASR). Our findings further demonstrate the need for strong guardrails and modules for agentic systems.
Yannis Belkhiter, Giulio Zizzo, Sergio Maffeis et al.· 0 citations
While prior red-teaming efforts have focused on eliciting harmful text outputs from large language models (LLMs), such approaches fail to capture agent-specific vulnerabilities that emerge through multi-step tool execution, particularly in rapidly growing ecosystems such as the Model Context Protocol (MCP). To address this gap, we propose a trajectory-aware evolutionary search method, T-MAP, which leverages execution trajectories to guide the discovery of adversarial prompts. Our approach enables the automatic generation of attacks that not only bypass safety guardrails but also reliably realize harmful objectives through actual tool interactions. Empirical evaluations across diverse MCP environments demonstrate that T-MAP substantially outperforms baselines in attack realization rate (ARR) and remains effective against frontier models, including GPT-5.2, Gemini-3-Pro, Qwen3.5, and GLM-5, thereby revealing previously underexplored vulnerabilities in autonomous LLM agents. Code is available at https://github.com/pwnhyo/T-MAP.
Hyomin Lee, Sangwoo Park, Yumin Choi et al.· 0 citations
Watermarking has emerged as a crucial technique for detecting and attributing content generated by large language models. While recent advancements have utilized watermark ensembles to enhance robustness, prevailing methods typically prioritize maximizing the strength of the watermark at every individual layer. In this work, we identify a critical limitation in this "stronger-is-better" approach: strong watermarks significantly reduce the entropy of the token distribution, which paradoxically weakens the effectiveness of watermarking in subsequent layers. We theoretically and empirically show that detectability is bounded by entropy and that watermark ensembles induce a monotonic decrease in both entropy and the expected green-list ratio across layers. To address this inherent trade-off, we propose a general framework that utilizes weaker single-layer watermarks to preserve the entropy required for effective multi-layer ensembling. Empirical evaluations demonstrate that this counter-intuitive strategy mitigates signal decay and consistently outperforms strong baselines in both detectability and robustness.
Ruibo Chen, Yihan Wu, Xuehao Cui et al.· 0 citations
