Related papers: Defensible Design for OpenClaw: Securing Autonomou…

Computer network defence is a complicated task that has necessitated a high degree of human involvement. However, with recent advancements in machine learning, fully autonomous network defence is becoming increasingly plausible. This paper…

This position paper argues that securing LLM agents requires first defining an end-to-end correctness property that specifies when an agent's execution faithfully reflects the user's intent. Modern LLM agents operate over an…

Agentic Al systems are increasingly deployed as personal assistants and are likely to become a common object of digital investigations. However, little is known about how their internal state and actions can be reconstructed during forensic…

Personalized computer-use agents are rapidly moving from expert communities into mainstream use. Unlike conventional chatbots, these systems can install skills, invoke tools, access private resources, and modify local environments on users'…

AI safety is still largely framed as alignment: training models to follow human preferences, safety policies, and normative constraints. That framing has improved the behavior of modern language models, but aligned behavior does not by…

Autonomous systems are emerging in many application domains. With the recent advancements in artificial intelligence and machine learning, sensor technology, perception algorithms and robotics, scenarios previously requiring strong human…

Recent advances in AI agents capable of solving complex, everyday tasks, from scheduling to customer service, have enabled deployment in real-world settings, but their possibilities for unsafe behavior demands rigorous evaluation. While…

Large Language Models (LLMs) are increasingly deployed as agentic systems that plan, memorize, and act in open-world environments. This shift brings new security problems: failures are no longer only unsafe text generation, but can become…

This chapter argues that the reliability of agentic and generative AI is chiefly an architectural property. We define agentic systems as goal-directed, tool-using decision makers operating in closed loops, and show how reliability emerges…

Modern open-world agents such as OpenClaw exhibit powerful cross-environment execution capabilities yet introduce broad new safety risk sources. Meanwhile, advanced frontier AI models drastically lower attack barriers, rendering current…

Debuggers are a popular reverse engineering and tampering tool. Self-debugging is an effective technique for applications to defend themselves against hostile debuggers. In penetration tests on state-of-the-art self-debugging, we observed…

As autonomous vehicle (AV) technology advances towards maturity, it becomes imperative to examine the security vulnerabilities within these cyber-physical systems. While conventional cyber-security concerns are often at the forefront of…

To counter software reverse engineering or tampering, software obfuscation tools can be used. However, such tools to a large degree hard-code how the obfuscations are deployed. They hence lack resilience and stealth in the face of many…

Large language models increasingly operate as autonomous agents that select and invoke tools from large registries. We identify a critical gap: when unauthorized tools are visible in an agent's context, models select them in adversarial…

Agentic language-model systems increasingly rely on mutable execution contexts, including files, memory, tools, skills, and auxiliary artifacts, creating security risks beyond explicit user prompts. This paper presents DeepTrap, an…

When large language model (LLM) systems interact with external data to perform complex tasks, a new attack, namely prompt injection, becomes a significant threat. By injecting instructions into the data accessed by the system, the attacker…

Current evaluations of tool-integrated LLM agents typically focus on end-to-end tool-usage evaluation while neglecting their stability. This limits their real-world applicability, as various internal or external factors can cause agents to…

Despite their growing adoption across domains, large language model (LLM)-powered agents face significant security risks from backdoor attacks during training and fine-tuning. These compromised agents can subsequently be manipulated to…

Deep reinforcement learning has emerged as a powerful tool for obtaining high-performance policies. However, the safety of these policies has been a long-standing issue. One promising paradigm to guarantee safety is a shield, which shields…

While large language model-based agents demonstrate great potential in collaborative tasks, their interactivity also introduces security vulnerabilities. In this paper, we propose and model group collusive attacks, a highly destructive…