PISmith

PIForge An Open Framework for RL-based Prompt Injection Red Teaming 💻 Code · 🤗 Models · 📜 Papers PIForge is the shared codebase for PISmith and Climbing the Hill. PISmith addresses the sparse reward problem in prompt injection red teaming, helping RL attackers explore and learn from rare successf...

GuardReasoner-VL

GuardReasoner-VL: Safeguarding VLMs via Reinforced Reasoning Yue Liu, Shengfang Zhai, Mingzhe Du Yulin Chen, Tri Cao, Hongcheng Gao, Cheng Wang Xinfeng Li, Kun Wang, Junfeng Fang, Jiaheng Zhang, Bryan Hooi 1National University of Singapore, 2Nanyang Technological University To enhance the safety ...

CyberBattleSim

CyberBattleSim April 8th, 2021: See the announcement on the Microsoft Security Blog. CyberBattleSim is an experimentation research platform to investigate the interaction of automated agents operating in a simulated abstract enterprise network environment. The simulation provides a high-level...

Pesidious

Malware Mutation using Deep Reinforcement Learning and GANs The purpose of the tool is to use artificial intelligence to mutate a malware PE32 only sample to bypass AI powered classifiers while keeping its functionality intact. In the past, notable work has been done in this domain with researche...

AutoPentest-DRL

AutoPentest-DRL: Automated Penetration Testing Using Deep Reinforcement Learning AutoPentest-DRL is an automated penetration testing framework based on Deep Reinforcement Learning DRL techniques. AutoPentest-DRL can determine the most appropriate attack path for a given logical network, and can...

StealthRL

StealthRL: Reinforcement Learning Paraphrase Attacks for Multi-Detector Evasion of AI-Text Detectors Paper arXiv Demo Model Hugging Face Benchmark Dataset Hugging Face Abstract AI-text detectors are increasingly used in high-stakes settings, yet their robustness to meaning-preserving adversarial...

Alignment of Autonomous Cyber Defense Agents Via Adaptive Policy Shaping

Cybersecurity operators increasingly encounter sophisticated and fast-paced attacks, making autonomous cyber defense ACD a critical approach for supporting timely and effective decision-making. Reinforcement learning RL-based ACD agents have the ability to autonomously observe network states and...

Black-Box Red-Teaming of Multi-Agent Systems Via Reinforcement Learning

Large language model LLM agents are increasingly deployed in multi-agent systems MAS to accomplish complex tasks. Prior black-box red-teaming attacks mainly focus on single agents, but we find that these methods are far less effective in MAS, where multiple sub-agents may not directly interact wi...

TRAP: Tree-Guided Reinforcement-Augmented Probing for Automated LLM Jailbreaking

Aligned large language models LLMs remain vulnerable to jailbreak prompts that circumvent their safety guardrails, making the identification of these vulnerabilities important for understanding and mitigating model weaknesses. Existing automated jailbreak methods explore prompt candidates or reus...

Reward Hacking As a Transmissible Strategy in Multi-Agent Coding Systems

Multi-agent systems are increasingly used to solve coding tasks by coordinating agents that share code and plans. These agents are often optimized with reinforcement learning RL using automated evaluators, but bypassing evaluation can earn reward without producing a genuine solution. Prior work h...

The Deceptive Bandit Problem: Exploratory Coupling and the Fragility of Multi-Agent Learning

Randomized exploration is central to bandit learning, multi-agent reinforcement learning, and zeroth-order policy search, yet its independence and privacy are usually only treated as technical assumptions. We show that these properties are critical for security purposes and demonstrate how an...

Ask the Expert: LLM-Guided Reinforcement Learning for Autonomous Cyber Defense

Policy-based reinforcement learning RL approaches have produced promising results for autonomous cyber defense; however, they are sample-inefficient in settings where defenders must respond under delayed, partial observations with actions from large action spaces. While large language models LLMs...

Understanding and Enhancing Backdoor Persistency in LLM Agent Post-Training

Developers can build LLM agents by adapting third-party models through benign post-training. We study a supply-chain threat in which an attacker supplies a model with a backdoor: hidden behavior that produces malicious outputs when a particular input pattern appears. Focusing on...

Learnt Attacks on Quantum Key Distribution under Channel Noise and Device Drift

Quantum key distribution QKD links are provisioned from security analyses of stationary channels, whereas the devices that determine the channel drift between recalibrations. Whether an eavesdropper who cannot alter the channel's own noise gains by following that drift has not been quantified...

Homomorphic Advantage Operator: Stabilizing Reinforcement Learning under Fully Homomorphic Encryption Constraints

Privacy-preserving machine learning presents significant deployment challenges on the cloud for intelligent systems with confidential data. Fully Homomorphic Encryption FHE offers a compelling solution for secure computation, preserving data confidentiality of cloud computations. However, applyin...

CVE-2026-103270

LightLLM through 1.2.0 mounts reinforcement learning control routes on the public HTTP API without authentication checks. Unauthenticated attackers can call endpoints like /pausegeneration, /abortrequest, /flushcache, and /initweightsupdategroup to disrupt inference operations and wedge workers o...

CVE-2026-103270: Missing Authentication for Critical Function

LightLLM through 1.2.0 mounts reinforcement learning control routes on the public HTTP API without authentication checks. Unauthenticated attackers can call endpoints like /pausegeneration, /abortrequest, /flushcache, and /initweightsupdategroup to disrupt inference operations and wedge workers o...

Towards Hierarchical Cyber Defense with Large Language Models: From Planning to Execution

An autonomous cyber defender trained with reinforcement learning RL is typically tied to the network on which it was trained, limiting its ability to generalize as network scale changes. Hierarchical RL reduces decision complexity by separating strategic targeting from tactical execution, but it...

SecureVibe: Making Vibe Coding More Secure

As vibe coding becomes increasingly capable and widespread, security vulnerabilities in even functionally correct solutions are a growing concern. When investigating functionally correct but insecure solutions, we find that the insecure agent is less than half as likely to conduct effective...

AI-Based Vulnerability Assessment Capability and Cyber Attack Graph Analysis

Cyber threats targeting mission-critical infrastructure are becoming more sophisticated while the barrier to launching attacks continues to fall. Traditional point solutions like antivirus and firewalls are reactive and fail to address the combinatorial complexity of modern attack surfaces. This...