Bypassing AI Control Protocols Via Agent-As-A-Proxy Attacks

As AI agents automate critical workloads, they remain vulnerable to indirect prompt injection IPI attacks. Current defenses rely on monitoring protocols that jointly evaluate an agent's Chain-of-Thought CoT and tool-use actions to ensure alignment with user intent. We demonstrate that these...

LlamaFirewall: an Open Source Guardrail System for Building Secure AI Agents

Large language models LLMs have evolved from simple chatbots into autonomous agents capable of performing complex tasks such as editing production code, orchestrating workflows, and taking higher-stakes actions based on untrusted inputs like webpages and emails. These capabilities introduce new...

UBUNTU-CVE-2023-52494

In the Linux kernel, the following vulnerability has been resolved: bus: mhi: host: Add alignment check for event ring read pointer Though we do check the event ring read pointer by "isvalidringptr" to make sure it is in the buffer range, but there is another risk the pointer may be not aligned...

CVE-2017-17856

kernel/bpf/verifier.c in the Linux kernel through 4.14.8 allows local users to cause a denial of service memory corruption or possibly have unspecified other impact by leveraging the lack of stack-pointer alignment enforcement...