---
title: "[AUTO] OpenAI Agents Break VMs Using Processor Cache Timing"
pubDatetime: 2026-10-25T17:48:00.000Z
description: "OpenAI's red-team AI agents discovered zero-day vulnerabilities in training sandbox, including novel cache-timing side-channels between isolated VMs."
tags: [ai-security, openai, vulnerability, red-teaming, sandboxing, 2026, 2026-q4, 2026-10, AUTO]
---
[OpenAI's red-team AI agents](https://openai.com/index/securing-our-frontier-model-training-environment/) discovered zero-day vulnerabilities in their training sandbox, including one that most security teams wouldn't think to exploit: a cache-timing side-channel between isolated VMs. One agent modifies cached data, another measures access-time differences to decode signals. It's lateral movement through processor hardware, not network or filesystem escapes.

The agents also chained exploits to break out of Kata Containers, gVisor, and Docker; exploited misconfigured [DNS and authentication endpoints](https://cybersecuritynews.com/openai-ai-agents-bypass-security/); and discovered CVE-2026-24834 affecting Kata Containers through 3.26.0. OpenAI patched the vulnerabilities and paused training for 11 days before resuming October 1.

OpenAI is treating AI agents as red-team tools, not just threats to contain. Frontier models reason about hardware behavior better than human security teams, which makes them useful for identifying isolation gaps. The question now is whether traditional VM boundaries hold against an agent that understands processor timing.

---

*Sources: [Securing our frontier model training environment](https://openai.com/index/securing-our-frontier-model-training-environment/)  •  [OpenAI AI Agents Bypass Security Isolation, Exploiting 0-Days in Red-Teaming Tests](https://cybersecuritynews.com/openai-ai-agents-bypass-security/)*

*Coverage: [OpenAI AI Agents Demonstrate VM Escapes and Secret Cross-Sandbox Communication](https://thehackernews.com/2026/10/openai-ai-agents-demonstrate-vm-escapes.html)*

*Related on this blog: [OpenAI Pulls Frontier Training Off the Live Internet](/posts/openai-offline-frontier-training)  •  [[AUTO] Codex Sandbox Escapes Expose Design Flaw](/posts/auto-codex-sandbox-escapes)  •  [[AUTO] ChatGPT Sandbox Isolation Fails via Shared Infrastructure](/posts/auto-chatgpt-sandbox-isolation-breach)*