---
type: "Evidence Item"
title: "Running Codex safely at OpenAI"
description: "How OpenAI runs Codex securely with sandboxing, approvals, network policies, and agent-native telemetry to support safe and compliant coding agent adoption."
resource: "https://openai.com/index/running-codex-safely"
tags: ["appendix-iii", "vendor", "openai"]
timestamp: "2026-05-08"
category: "vendor"
publisher: "OpenAI"
cope_score: 74
confidence: 0.9
---

# Running Codex safely at OpenAI

# Claim

How OpenAI runs Codex securely with sandboxing, approvals, network policies, and agent-native telemetry to support safe and compliant coding agent adoption.

# Relevance

Appendix III, section two: vendor threshold and platform capability evidence

# Oracle Verdict

This is a lower-to-mid strength vendor signal for the capability register. It does not prove displacement on its own, but it records another platform step that can later show up as workflow automation, procurement change, or organisational dependency.

# Metadata

* Publisher: OpenAI
* Category: vendor
* Sector: Software engineering
* Capability: Autonomous software engineering and computer-use agents
* Cope score: 74
* Confidence: 0.9

# Related Concepts

* [Live evidence index](index.md)
* [Thesis](../thesis.md)

# Citations

[1] [Running Codex safely at OpenAI](https://openai.com/index/running-codex-safely)