> ## Content Index
> Fetch the complete content index at: https://www.agentsecbrief.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# What is local privilege escalation?
- URL: https://www.agentsecbrief.com/what-is-local-privilege-escalation/
- Published: 2026-11-06T16:47:57.000Z
- Updated: 2026-11-06T16:47:57.000Z
- Description: 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.
- Author: Angela Milash
- Tags: Sandboxing & Runtime, Glossary

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](https://openai.com/index/a-thousand-fewer-paths-to-root/?ref=agentsecbrief.com) 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](https://www.agentsecbrief.com/checklist/).

## Related terms

Sandbox escape, exploit chain, least privilege.

## Sources

- [OpenAI, November 2026](https://openai.com/index/a-thousand-fewer-paths-to-root/?ref=agentsecbrief.com)
- [Red Hat RHSB-2026-016](https://access.redhat.com/security/vulnerabilities/RHSB-2026-016?ref=agentsecbrief.com)

*From AgentSec Brief, the daily brief on securing agentic AI. Subscribe free at* [*https://www.agentsecbrief.com*](https://www.agentsecbrief.com/)