What is local privilege escalation?
Local privilege escalation is when code that already runs on a computer with limited rights finds a flaw that gives it full administrator (root) control.
Local privilege escalation is when code that already runs on a computer with limited rights finds a flaw that gives it full administrator (root) control.
Why it matters
Many AI agent sandboxes run untrusted code as a low-privilege user in a container that shares the host's operating system kernel. A kernel privilege escalation can turn that code into root on the host, past the sandbox, and AI is making such exploits cheaper to produce.
Example
On 4-5 November 2026, OpenAI reported that its agent pipeline turned 2,308 Linux kernel findings into 948 working privilege-escalation exploits in about 30 hours (self-reported).
How to defend against it
- Run untrusted agent code in a microVM or a sandbox with its own kernel, not only a shared-kernel container.
- Patch kernels quickly on hosts that run agent workloads.
- Run agent code as a non-root user with the fewest system calls allowed.
- Alert on processes from sandbox users that gain root. See the AgentSec Brief checklist.
Related terms
Sandbox escape, exploit chain, least privilege.
Sources
From AgentSec Brief, the daily brief on securing agentic AI. Subscribe free at https://www.agentsecbrief.com