NewEvery arXiv paper, its researchers & institutions — mapped.
the archive

cs.CR

34.5k results
cs.CR2026

AnchorMark: Robust Diffusion Watermarking via Latent-Space Rotation Synchrony

Yuqi Qian, Yun Cao, Haocheng Fu +3

The paper proposes AnchorMark, a training‑free inversion‑based watermarking method for diffusion models that uses a latent‑space rotation synchrony property to embed a central anch…

#watermarking#diffusion models#latent space#rotation robustness
cs.CR2026

Temporal Poisoning: Clean-Label Backdoors via Event Redistribution in SNNs

Roberto Riaño, Gorka Abad, Stjepan Picek +1

The paper introduces a clean‑label backdoor attack for spiking neural networks that subtly reshapes the timing of events in target‑class training streams, achieving near‑perfect at…

#spiking neural networks#backdoor attacks#clean-label poisoning#temporal manipulation
cs.CR2026

Piggybacking on Perception: Stealthy Concurrent Audio Prompt Injections against Multimodal LLM Agents

Mingxiao Liu, Yitong Li, Haoren Zhao +6

The paper studies stealthy audio prompt injection attacks that hide malicious instructions within normal speech to hijack multimodal LLM agents, introduces a benchmark (AudioAgentS…

#audio injection#multimodal llm#prompt injection#adversarial attacks
cs.CR2026

Strategy Phasing of Cyber Attacks on Digital Substations

Akila Herath, Chen-Ching Liu, Junho Hong +1

The paper introduces SubCASP, a Hidden Markov Model approach that combines IDS logs to identify and predict the phases of cyber attacks on IEC 61850‑based digital substations, usin…

#digital substations#iec 61850#intrusion detection systems#attack phase inference
cs.CR2026

CHARGE: Leveraging CWE Hierarchies for Hardware Security SystemVerilog Assertion Generation

Xiao Tan, Cynthia Sturton

CHARGE is an automated framework that uses CWE hierarchies and large language models to generate SystemVerilog assertions for unverified RTL modules, enabling security property inf…

#systemverilog assertions#cwe hierarchy#large language models#hardware security verification
cs.CR2026

Agent Harness Distillation: Inference-Time Harness Extraction and Exploitation in Autonomous Multi-Agent Systems

Yu Cui, Wuli Yang, Yirui Shi +4

The paper presents Agent Harness Distillation (AHD), a method for extracting and replicating inference-time coordination mechanisms (harnesses) from autonomous multi-agent systems…

#inference-time harness#autonomous multi-agent systems#llm security#ip leakage
cs.CR2026

Safeguards Based on Copyable Context Cannot Provide Reliable Safety for LLMs

Pingyu Wu, Lingyao Zhu, Weiming Zhang +1

The paper shows that safeguards for large language models that rely only on copyable context cannot guarantee reliable safety for dual‑use tasks, and proposes adding hard‑to‑copy t…

#large language models#ai safety#dual-use#access control
cs.CR2026

Adaptive Security at the Edge for 6G-Enabled Healthcare IoT

Ijaz Ahmad, Erkki Harjula

The paper introduces NANOEDGEGUARD, a kernel‑plane controller that adaptively enforces multi‑tier rate limits at edge gateways for healthcare IoT, improving alarm latency and reduc…

#edge computing#healthcare iot#adaptive rate control#kernel-plane enforcement
cs.CR2026

Distributed Point Functions and Function Secret Sharing

Elette Boyle, Niv Gilboa, Yuval Ishai +1

The paper surveys the cryptographic primitive of distributed point functions (DPFs) and its extension to function secret sharing (FSS), covering definitions, constructions, and a r…

#distributed point functions#function secret sharing#private information retrieval#anonymous messaging
cs.CR2026

Benign on Label, Malicious by Design: Clean-Label Dormant-to-Activated Backdoor via Machine Unlearning with Removable Camouflage

Dongdong Zhao, Can Li, Xiang Yao +3

The paper proposes a clean‑label backdoor attack that stays dormant during training and becomes active only after specific camouflage samples are removed via machine unlearning, us…

#backdoor attacks#machine unlearning#clean-label#adversarial machine learning
cs.CR2026

Secure Aggregation for Privacy-Preserving Federated Learning on Clinical EEG Data

Pouya Rajabi, Mohsen Toorani

The paper proposes a federated learning framework for clinical EEG data that uses masking‑based secure aggregation and related cryptographic techniques to protect individual model…

#federated learning#secure aggregation#privacy preservation#clinical eeg
cs.CR2026

Implementing Homomorphic Encryption-Based Logic Locking in System-on-Chip Designs

Ye Ziyang, Makoto Ikeda

The paper proposes a logic locking scheme for RISC‑V system‑on‑chip designs that uses the binary Ring‑LWE homomorphic encryption algorithm to protect privileged logic paths without…

#logic locking#homomorphic encryption#ring-lwe#risc-v
cs.CR2026

Checking Information Flow in Cloud-based IoT Access Control Policies (Extended Version)

Lorenzo Ceragioli, Letterio Galletta, Edoardo Lunati

The paper presents a method to detect unwanted information flow in cloud‑based IoT access control policies by modeling AWS IoT Core components and using an SMT‑based analysis tool…

#iot security#access control#information flow#cloud computing
cs.CR2026

Open Security Benchmark: Towards Autonomous Enterprise Cyber Defense

Gal Engelberg, Michael Arenzon, Leon Goldberg

The paper introduces the Open Security Benchmark (OSB), a framework that provides a frozen, holistic enterprise security dataset and evaluation tools for testing autonomous AI agen…

#autonomous cyber defense#security posture management#benchmarking#enterprise security
cs.CR2026

InkShield: Writing Style Protection Against Unauthorized Handwriting Mimicry

Jian Xiong, Wenbo Jiang, Zihan Wang +4

InkShield introduces a proactive defense that adds subtle, stroke‑confined perturbations to handwritten reference images, making it harder for handwriting generators to mimic a wri…

#handwriting authentication#adversarial perturbations#style protection#biometric security
cs.CR2026

ToxScreen: Detecting Whether an LLM Has Been Poisoned

Anthony Hughes, Nicole Xing, Collin Francel +2

The paper introduces ToxScreen, a benchmark of backdoored large language models, and evaluates methods for recovering hidden triggers under realistic defender constraints, finding…

#large language models#backdoor detection#adversarial attacks#model security
cs.CR2026

QUIC-TRIP: A Triple-Redundant Journey Toward Secure Substation Communications

Jorge David de Hoz Diego, Ioannis Zografopoulos, Anca Jurcut

The paper presents QUIC-TRIP, a transport‑layer security method that adds low‑latency, triple‑redundant protection to substation communication protocols like R‑GOOSE without modify…

#substation communication#secure transport#multipath redundancy#low-latency security
cs.CR2026

Recover, Decode, Reguard: Guard-Agnostic Defense Amplification againstEncoded VLM Jailbreaks

Haoyu Zhang, Zhuoxi Wang, Shibo Zheng +4

The paper proposes a guard‑agnostic recovery‑and‑decode module that transcribes encoded or visual text into plain language before applying existing safety classifiers for vision‑la…

#vision-language models#safety guards#jailbreak attacks#recovery decoding
cs.CR2026

Guarding Organizations Against Malware Risk: A Novel Graph-Based Malware Detection Method

Yinan Gao, Jiarong Xu, Xiaohang Zhao +1

The paper introduces MalGuard, a graph‑based malware detection system that groups basic blocks into operational roles and learns expressive program‑graph representations to improve…

#malware detection#graph neural networks#program analysis#operational role identification
cs.CR2026

Graph Is the Verifier: Agentic Reinforcement Learning for Interprocedural Vulnerability Detection

Yikun Li, Ting Zhang, Jiakun Liu +9

The paper presents VulAgentRL, an agentic reinforcement learning framework that leverages code property graphs to collect interprocedural evidence and verify its own reasoning for…

#vulnerability detection#interprocedural analysis#reinforcement learning#code property graph
cs.CR2026

Not In My Git Yard: Catching Backdoors at Commit and Release Time

Dimitri Kokkonis, Michaël Marcozzi, Stefano Zacchiroli

The paper introduces Lily, a tool that automatically detects hidden backdoors in open‑source code during commit and release stages by combining CI‑compatible fuzzing with code‑chan…

#backdoor detection#continuous integration#fuzz testing#supply chain security
cs.CR2026

FARI: Robust One-Step Inversion for Watermarking in Diffusion Models

Jindong Yang, Han Fang, Weiming Zhang +2

The paper introduces FARI, a fast one-step inversion method combined with lightweight adversarial LoRA fine-tuning to robustly extract watermarks from diffusion-generated images, a…

#diffusion models#watermarking#image inversion#adversarial training
cs.CR2026

Verifiable Random Sampling

Yeoh Wei Zhu, Soorya Rethinasamy, Anthony Alexiades Armenakas +3

The paper introduces verifiable random sampling (VRS), a cryptographic primitive that uses random quantum circuit sampling to produce publicly verifiable, fresh random samples, add…

#verifiable random sampling#quantum circuit sampling#public verifiability#constructive cryptography
cs.CR2026

SecRespond: Benchmarking AI Agents for Real-World Post-Compromise Incident Response

Lehan Wang, Boli Chen, Ruixue Ding +7

The paper presents SecRespond, a benchmark that evaluates large language model agents on post-compromise incident‑response tasks using forensic disk snapshots, alerts, and vulnerab…

#incident response#large language models#benchmark#post-compromise