Running Codex safely at OpenAI
📰 OpenAI News
Learn how OpenAI runs Codex safely with controls, boundaries, and telemetry to govern coding agents in real workflows, and apply these principles to your own AI deployments
Action Steps
- Configure sandboxing and approvals for your AI agents to ensure they operate within clear technical boundaries
- Implement network access controls to restrict what systems your AI agents can interact with
- Set up identity and credentials management to govern agent access to sensitive resources
- Establish rules and managed configs to constrain agent execution and preserve telemetry
- Monitor agent-native telemetry and audit trails to understand and explain agent behavior
Who Needs to Know This
Developers, security teams, and organizations using AI systems like Codex can benefit from understanding how to deploy AI safely and effectively, with clear boundaries and controls
Key Insight
💡 Clear boundaries, controls, and telemetry are crucial for safe and effective AI deployment
Share This
🚀 Learn how to run Codex safely with controls, boundaries, and telemetry! 🤖
Key Takeaways
Learn how OpenAI runs Codex safely with controls, boundaries, and telemetry to govern coding agents in real workflows, and apply these principles to your own AI deployments
Full Article
Title: Running Codex safely at OpenAI | OpenAI
URL Source: https://openai.com/index/running-codex-safely
Markdown Content:
[Skip to main content](https://openai.com/index/running-codex-safely#main)
[](https://openai.com/)
* [Research](https://openai.com/research/index/)
* Products
* [Business](https://openai.com/business/)
* [Developers](https://openai.com/api/)
* [Company](https://openai.com/about/)
* [Foundation(opens in a new window)](https://openaifoundation.org/)
Log in[Try ChatGPT(opens in a new window)](https://chatgpt.com/)
* Research
* Products
* Business
* Developers
* Company
* [Foundation(opens in a new window)](https://openaifoundation.org/)
[Try ChatGPT(opens in a new window)](https://chatgpt.com/)Login
OpenAI
May 8, 2026
[Security](https://openai.com/news/security/)[Safety](https://openai.com/news/safety-alignment/)
# Running Codex safely at OpenAI
A look at the controls, boundaries, and telemetry OpenAI uses to govern coding agents in real workflows.
Loading…
[Audio 1](https://openai.com/index/running-codex-safely)
Share
Controlling how Codex operates
* [Controlling how Codex operates](https://openai.com/index/running-codex-safely#controlling-how-codex-operates)
* [Sandboxing and approvals](https://openai.com/index/running-codex-safely#sandboxing-and-approvals)
* [Network access](https://openai.com/index/running-codex-safely#network-access)
* [Identity and credentials](https://openai.com/index/running-codex-safely#identity-and-credentials)
* [Rules](https://openai.com/index/running-codex-safely#rules)
* [Managed configs](https://openai.com/index/running-codex-safely#managed-configs)
* [Agent-native telemetry and audit trails](https://openai.com/index/running-codex-safely#agent-native-telemetry-and-audit-trails)
* [Looking ahead](https://openai.com/index/running-codex-safely#looking-ahead)
Table of contents
* [Controlling how Codex operates](https://openai.com/index/running-codex-safely#controlling-how-codex-operates)
* [Sandboxing and approvals](https://openai.com/index/running-codex-safely#sandboxing-and-approvals)
* [Network access](https://openai.com/index/running-codex-safely#network-access)
* [Identity and credentials](https://openai.com/index/running-codex-safely#identity-and-credentials)
* [Rules](https://openai.com/index/running-codex-safely#rules)
* [Managed configs](https://openai.com/index/running-codex-safely#managed-configs)
* [Agent-native telemetry and audit trails](https://openai.com/index/running-codex-safely#agent-native-telemetry-and-audit-trails)
* [Looking ahead](https://openai.com/index/running-codex-safely#looking-ahead)
As AI systems become more capable, they increasingly act on behalf of users. Coding agents can autonomously review repositories, run commands, and interact with development tools. These are tasks that previously required direct human execution.
With Codex, we’ve designed these capabilities alongside the controls organizations need for safe deployment. Security teams need ways to govern how agents operate: what they can access, when human approval is required, which systems they can interact with, and what telemetry exists to explain their behavior.
At OpenAI, we deploy Codex with a few clear goals: keep the agent inside clear technical boundaries, let developers move quickly on low-risk actions, and make higher-risk actions explicit. We also preserve agent-native telemetry so we can understand and audit what the agent did. In practice, that means managed configuration, constrained execution, network policies, and agent-native logs.
## Controlling how Codex operates
We deploy Codex with a simple principle that it should be productive inside a bounded environment, low-risk everyday actions should be frictionless, and higher-risk actions should stop for review.
### Sandboxing and approvals
Approvals and sandboxing work together. The sand
URL Source: https://openai.com/index/running-codex-safely
Markdown Content:
[Skip to main content](https://openai.com/index/running-codex-safely#main)
[](https://openai.com/)
* [Research](https://openai.com/research/index/)
* Products
* [Business](https://openai.com/business/)
* [Developers](https://openai.com/api/)
* [Company](https://openai.com/about/)
* [Foundation(opens in a new window)](https://openaifoundation.org/)
Log in[Try ChatGPT(opens in a new window)](https://chatgpt.com/)
* Research
* Products
* Business
* Developers
* Company
* [Foundation(opens in a new window)](https://openaifoundation.org/)
[Try ChatGPT(opens in a new window)](https://chatgpt.com/)Login
OpenAI
May 8, 2026
[Security](https://openai.com/news/security/)[Safety](https://openai.com/news/safety-alignment/)
# Running Codex safely at OpenAI
A look at the controls, boundaries, and telemetry OpenAI uses to govern coding agents in real workflows.
Loading…
[Audio 1](https://openai.com/index/running-codex-safely)
Share
Controlling how Codex operates
* [Controlling how Codex operates](https://openai.com/index/running-codex-safely#controlling-how-codex-operates)
* [Sandboxing and approvals](https://openai.com/index/running-codex-safely#sandboxing-and-approvals)
* [Network access](https://openai.com/index/running-codex-safely#network-access)
* [Identity and credentials](https://openai.com/index/running-codex-safely#identity-and-credentials)
* [Rules](https://openai.com/index/running-codex-safely#rules)
* [Managed configs](https://openai.com/index/running-codex-safely#managed-configs)
* [Agent-native telemetry and audit trails](https://openai.com/index/running-codex-safely#agent-native-telemetry-and-audit-trails)
* [Looking ahead](https://openai.com/index/running-codex-safely#looking-ahead)
Table of contents
* [Controlling how Codex operates](https://openai.com/index/running-codex-safely#controlling-how-codex-operates)
* [Sandboxing and approvals](https://openai.com/index/running-codex-safely#sandboxing-and-approvals)
* [Network access](https://openai.com/index/running-codex-safely#network-access)
* [Identity and credentials](https://openai.com/index/running-codex-safely#identity-and-credentials)
* [Rules](https://openai.com/index/running-codex-safely#rules)
* [Managed configs](https://openai.com/index/running-codex-safely#managed-configs)
* [Agent-native telemetry and audit trails](https://openai.com/index/running-codex-safely#agent-native-telemetry-and-audit-trails)
* [Looking ahead](https://openai.com/index/running-codex-safely#looking-ahead)
As AI systems become more capable, they increasingly act on behalf of users. Coding agents can autonomously review repositories, run commands, and interact with development tools. These are tasks that previously required direct human execution.
With Codex, we’ve designed these capabilities alongside the controls organizations need for safe deployment. Security teams need ways to govern how agents operate: what they can access, when human approval is required, which systems they can interact with, and what telemetry exists to explain their behavior.
At OpenAI, we deploy Codex with a few clear goals: keep the agent inside clear technical boundaries, let developers move quickly on low-risk actions, and make higher-risk actions explicit. We also preserve agent-native telemetry so we can understand and audit what the agent did. In practice, that means managed configuration, constrained execution, network policies, and agent-native logs.
## Controlling how Codex operates
We deploy Codex with a simple principle that it should be productive inside a bounded environment, low-risk everyday actions should be frictionless, and higher-risk actions should stop for review.
### Sandboxing and approvals
Approvals and sandboxing work together. The sand
DeepCamp AI