Related papers: Sandlock: Confining AI Agent Code with Unprivilege…

Data analyses in the life sciences are moving from tools run on a personal computer to services run on large computing platforms. This creates a need to package tools and dependencies for easy installation, configuration and deployment on…

As AI agents move from demos into enterprise deployments, their failure modes become consequential: a misinterpreted tool argument can corrupt production data, a silent reasoning error can go undetected until damage is done, and outputs…

Linux Containers, such as those managed by Docker, are an increasingly popular way to package and deploy complex applications. However, the fundamental security primitive of network access control for a distributed microservice deployment…

We introduce a new mechanism to securely extend systems data collection software with potentially untrusted third-party code. Unlike existing tools which run extension modules or plugins directly inside the monitored endpoint (the guest),…

Modern AI-powered Integrated Development Environments (AI-IDEs) are increasingly defined by an Agent-centric architecture, where an LLM-powered Agent is deeply integrated to autonomously execute complex tasks. This tight integration,…

Frontier AI developers are relying on layers of safeguards to protect against catastrophic misuse of AI systems. Anthropic and OpenAI guard their latest Opus 4 model and GPT-5 models using such defense pipelines, and other frontier…

Large Language Model (LLM) agents show considerable promise for automating complex tasks using contextual reasoning; however, interactions involving multiple agents and the system's susceptibility to prompt injection and other forms of…

In this paper we introduce Timeloops a novel technique for automatically learning system call filtering policies for containerized microservices applications. At run-time, Timeloops automatically learns which system calls a program should…

AI agents are beginning to interact with each other directly and across internet platforms and physical environments, creating security challenges beyond traditional cybersecurity and AI safety frameworks. Free-form protocols are essential…

AI agents increasingly call external tools (file system, network, APIs) through the Model Context Protocol (MCP). These tool calls are the agent's syscalls -- privileged operations with side effects on shared state -- yet today's safety…

Attacks on the microarchitecture of modern processors have become a practical threat to security and privacy in desktop and cloud computing. Recently, cache attacks have successfully been demonstrated on ARM based mobile devices, suggesting…

AI agents are increasingly autonomous in their interactions with human users and tools, leading to increased interactional safety risks. We present HAICOSYSTEM, a framework examining AI agent safety within diverse and complex social…

In a process algebra with hiding and recursion it is possible to create processes which compute internally without ever communicating with their environment. Such processes are said to diverge or livelock. In this paper we show how it is…

Enterprises increasingly rely on cloud-based applications to process highly sensitive data artifacts. Although cloud adoption improves agility and scalability, it also introduces new security challenges such as expanded attack surfaces, a…

Autonomous AI agents built on open-source runtimes such as OpenClaw expose every available tool to every session by default, regardless of the task. A summarization task receives the same shell execution, subagent spawning, and credential…

The A2AS framework is introduced as a security layer for AI agents and LLM-powered applications, similar to how HTTPS secures HTTP. A2AS enforces certified behavior, activates model self-defense, and ensures context window integrity. It…

Quest-V is a system we have been developing from the ground up, with objectives focusing on safety, predictability and efficiency. It is designed to work on emerging multicore processors with hardware virtualization support. Quest-V is…

Recent progress in (Large) Language Models (LMs) has enabled the development of autonomous LM-based agents capable of executing complex tasks with minimal supervision. These agents have started to be integrated into systems with significant…

AI-based systems, currently driven largely by LLMs and tool-using agentic harnesses, are increasingly discussed as a possible threat to software engineering. Foundation models get stronger, agents can plan and act across multiple steps, and…

Linux Seccomp is widely used by the program developers and the system maintainers to secure the operating systems, which can block unused syscalls for different applications and containers to shrink the attack surface of the operating…