cs.CR Computer Science
Cryptography and Security papers on arXiv
Follow security analysis, privacy, and cryptographic methods in cs.CR. Choose a threat model, system, or primitive so that your feed reflects the kind of security work you can use.
2,303 papers indexed
Follow the whole category with the official cs.CR RSS feed, or use a personal filter to select keywords and authors. See the arXiv RSS feed guide for setup and the category guide for primary and cross-listed subjects.
Filter ideas for Security & Cryptography
How filters workA broad security keyword can pull in background discussion. Combine the system being studied with the attack or defense, and use author filters when following a particular research group.
- Try filter
Differential privacy
Add a data setting or learning method to separate theoretical guarantees from application studies.
Category cs.CR + abstract contains “differential privacy” - Try filter
Post-quantum cryptography
Try an OR group with lattice if relevant work uses the primitive name rather than post-quantum terminology.
Category cs.CR + abstract contains “post-quantum”
Examples are starting points. Preview them before saving; keyword matches do not assess a paper’s quality. Recent papers below use the primary category; saved category filters also match cross-listed categories.
LATEST SUBMISSIONS
Recent Cryptography and Security papers
Mikiyas Alemayehu, Mohamed Chahine Ghanem, Kevin Curran, Hamza Kheddar, Ioanna Dionysiou, Elhadj Benkhelifa, J. J. Garcia-Luna-Aceves
Abstract
This paper presents a hybrid neural-symbolic detector of distributed denial-of-service (DDoS) attacks on the availability of operational technology (OT) networks, designed for their resource-constrained gateways, and an evaluation matched to deployment. A compact gated recurrent unit (GRU) is fused at score level with a shallow, readable decision tree, and weight and threshold are selected on validation data under an explicit false-positive budget, whose attainment in finite samples is analysed. The evaluation on CIC-DDoS2019, Edge-IIoTset and CICIoT23 uses contiguous or capture-held-out partitions, dependence-aware intervals, and precision and alert load at operational prevalence. Over ten seeds at a 1% budget, the hybrid's false-negative rate on CIC-DDoS2019 is 0.08%, against 0.26% for the GRU and 0.17% for the tree. With a day of mostly unseen attack families held out, the tree alone carries the decision. Audits bound these claims: Edge-IIoTset is saturated by duplicates, no benchmark as used provides interleaved traffic in capture order, and an unconstrained tree is more accurate on two of them, so the case for the hybrid rests on short, stable rules and a measured footprint. On a Siemens SIMATIC IoT2050 gateway, tested with Modbus/TCP and MQTT floods, the INT8 model needs 2.9 ms per window, 31 MB of memory and 3.5 mJ per inference.
Pingyu Wu, Weiming Zhang, Nenghai Yu
Abstract
Large language models now help professionals in fields such as cybersecurity and biology, yet the same answers would help an attacker. Deployers rely on misuse safeguards to keep such answers from attackers. We find that these safeguards share a high-risk vulnerability: they decide whether to release an answer from what users write, including the professional identity or purpose they claim, and anyone can write such a claim. Threat reports from model vendors and security firms show that attackers who fabricate a professional identity or purpose already exploit this vulnerability to cause serious harm. The vulnerability is also widespread across the state of the art: every defense proposed in 129 papers from the latest editions of 12 top venues has it. Safeguard research studies the wrong object: it treats harm as a property of the text of a request, although the same text can come from a professional or an attacker. Current safeguard benchmarks therefore cannot reveal the vulnerability: a safeguard can score high on them and still give dangerous answers to attackers. Because the vulnerability lies in what users write, no improvement in how safeguards judge that writing can close it. Under the patch we propose, large language models can help professionals without giving dangerous capabilities to anyone who merely claims to be one.
Pedro Beltrán-López, Raul Gonzalez Jimenez, Dario Maddaloni, Manuel Gil Pérez, Pantaleone Nespoli
Abstract
Large Language Models (LLMs) enable honeypots to generate responses and deceptive artefacts on demand, yet existing approaches evaluate generated content one reply at a time and leave open whether artefacts that jointly describe a coherent event retain an attacker better than unrelated ones. This question becomes pressing as autonomous LLM-driven offensive agents emerge as realistic adversaries. This paper proposes narrative deception, a reactive strategy in which the defender maps each observed attacker command to a MITRE ATT&CK technique, selects a compatible scenario from a curated catalogue based on MITRE Engage, and deploys mutually referential artefacts (files, processes, services, and logs) on an emulated Ubuntu host while the intrusion is under way. An LLM-powered defender and an autonomous LLM attacker interact through a real SSH session, and the attacker's termination policy distinguishes abandonment due to suspected deception from abandonment due to loss of interest. A campaign of 1,150 sessions compares four conditions of increasing sophistication (Static, Same-TTP, Multi-TTP, and Chain) across four open-weight attacker models, four defender models, and three attacker skill levels. Engagement increases significantly at every step. With the same artefact budget per reaction, coherent scenarios raise the median session length of advanced attackers from 40 to 165 commands and reduce their abandonment from 86% to 52%. The effect grows with attacker skill, holds across all sixteen model pairs, and comes at comparable inference cost. Nearly all sessions ended through loss of interest rather than detection, which suggests that deception should sustain meaningful findings, not only emulator fidelity.
Yujun Wang, Yilun Liu, Jinhe Bi, Krishna Kanth Nakka, Xuebing Zhou, Hinrich Sch{ü}tze, Yunpu Ma
Abstract
Tool-using language-model agents must act on information from untrusted emails, webpages, and documents. This creates a central security challenge: malicious instructions and legitimate task directives can arrive through the same interaction channel. Existing defenses are largely hand-designed and static, making further improvements dependent on repeated manual redesign. We introduce Evolution Under Fire, a staged framework that couples safety-harness evolution with adaptive stressor evolution. Utility-constrained, Pareto-scheduled tree search evolves complete executable safety harnesses from execution evidence while keeping the task model, benchmark, tools, and evaluator fixed. Adaptive stressor evolution then discovers utility-preserving attacks that guide a subsequent stage of harness evolution. On AgentDyn with GPT-4o-mini, an evolved harness reduces attack success from 50.4% to 3.0%, improves clean and attacked utility over the initial harness, and offers a stronger security-utility trade-off than reproduced defenses. On Agent-SafetyBench with Qwen2.5-7B-Instruct, harness evolution reduces attack success from 82.6% to 62.0% while retaining helpfulness and runtime validity. Evolved harnesses also transfer without further evolution to Qwen3.5-35B-A3B, reducing attack success from 32.1% to 0.9% while improving attacked-task utility.
Pingyu Wu, Weiming Zhang, Nenghai Yu
Abstract
Safeguards for large language model services are typically evaluated by their attack-blocking rates on test suites. Deployment decisions also depend on how much harmful assistance attackers can still obtain from the system, which these rates do not measure. To determine what published evaluations establish about deployment safety, we develop a framework that translates reported results into the minimum and maximum harmful assistance that can remain under a specified outcome and threat model. For safeguards that block attacks through a check, we show that the most an attacker can obtain depends on three properties: whether attack paths pass through the check, its miss rate against every attacker in the threat model, and how much assistance remains after a block. We apply this framework to 85 safeguard evaluations from 52 papers. Of these, only five argue that every attack path passes through the check, and four provide evidence on what remains after a block. These evidence gaps have practical consequences: in NeMo Guardrails' default streaming configuration, the check blocked 94% of password leaks in our test, yet the client received the password in every leak, blocked or not. We identify which of the three properties each safeguard category's evaluations omit, and what to measure for each of four ways attacks bypass a check. Deployments that require these measurements before launch can choose and configure safeguards that limit what attackers obtain from the deployed system.
Frequently asked questions
How do I get alerts for new Cryptography and Security papers on arXiv?
Create a filter for the cs.CR category — add keywords or authors to narrow it further if you like. Matching papers are delivered to a personal RSS feed, and optionally by immediate, daily, or weekly email digest.
Where does this Cryptography and Security paper data come from?
Paper metadata comes directly from arXiv through its official OAI-PMH feed, checked roughly every 15 minutes for new submissions in cs.CR and every other arXiv category.