
Secure Codex across your organization.
Discover Codex across the enterprise, in the terminal, the IDE and the desktop app, control what it reaches, enforce security policies and guardrails, and prevent data leakage.
- SOC 2 Type II
- SSO, SCIM & SIEM
Trusted by security teams · 5M+ agents protected
OpenAI Codex security in the terminal, the IDE and the desktop app
- See every Codex install and session, and who runs it.
- Block a risky Codex action before it runs, and mask a secret before it leaves.
- Govern the MCP servers and models Codex may use, by team.
What is Autonomous?
Codex enterprise security and governance: one platform to govern Codex usage at work, and the other supported AI agents.
It finds every Codex install and who runs it, the Codex CLI, the IDE extension and the desktop app, with the MCP servers and credentials it uses, through a native endpoint agent, and monitors Codex agent activity. It keeps an audit trail of activity on Codex's enterprise accounts through OpenAI's Compliance API. It discovers the shadow AI agents beside it. It scores every MCP server, skill and AI app for risk. It enforces what is approved, restricted or prohibited, and stops unsafe actions inline with AI guardrails and deterministic DLP. It secures over 60 of the most popular AI agents, from the endpoint to the browser and SaaS.

- SSO / SCIM
- SIEM integration
Codex security, in the terminal, the IDE and the desktop app.
OpenAI Codex works in the terminal, the IDE and the desktop app, with the developer's own access. Codex enterprise security covers all three.
What you see: every Codex install and session
Every Codex install with its MCP servers and credentials, and each session scored for risk. It keeps an audit trail of activity on Codex's enterprise accounts through OpenAI's Compliance API.
Codex guardrails
Guardrails check Codex's prompts, tool calls and tool results through its hooks, and data from a source you marked sensitive can't leave for an untrusted destination.
Codex governance, team by team
Allow, restrict or block the MCP servers, tools and models Codex may use, by team. Session and enforcement events stream to your SIEM.
OpenAI Codex security in a 14-day free trial
14 days, full access, on your own Codex rollout, no credit card. You leave with an inventory of Codex and your other AI agents, risk findings and tested policies.
What security leaders tell us.
“Autonomous Security gave us continuous, fleet-wide visibility into how Claude, Cursor, and Codex are actually being used, and helped us identify security exposures long after initial deployment.”
Matthew K. Sharp
CISO
Testimonial 1 of 4: Matthew K. Sharp, CISO, Xactly
Questions about Codex security.
Can we extend the free trial?
Yes. If you feel you need more time to get to know our product, just say the word and we'll extend your trial.
How is Autonomous priced?
A fixed, tiered base subscription, based on the number of endpoints our platform monitors. Customers can also tailor it to their exact needs with à la carte options. Contact us to get a detailed quote.
Can it stop a risky Codex action before it runs?
Yes. The Codex CLI and Codex Desktop expose hooks, so guardrails act inline: a finding set to Block denies the action, and one set to Redact masks a secret before it leaves.
Can we keep Codex on our own ChatGPT workspace?
Yes. Codex signs in through ChatGPT and reports its Workspace ID. Allow your workspace, and any other account, personal ones included, is blocked at the prompt once the policy is enforced.
How long does it take to get running?
About 10 minutes to integrate. The endpoint agent runs on macOS, Windows, Linux and containers, in the cloud too: install it with one command, or silently through your MDM (Microsoft Intune, Jamf Pro, Iru (Kandji), JumpCloud, Fleet or any other MDM). Its first scan inventories a machine's AI agents within a few minutes.
Will it disrupt our developers or block their work?
No, not unless you choose to. Every new policy and guardrail starts off or in monitor mode, where activity is evaluated and logged but nothing is blocked. You move a policy to enforce when its findings look right, for a pilot group first.
What data does it collect, and who can see it?
By default it collects the AI agents' prompts, tool calls, responses and session metadata, for audit and investigation, and only people whose role allows it see the content. You can set prompts and tool calls to keep only the verdict and metadata, only flagged activity, or nothing, and turn the other kinds off; a block is always recorded. Personal accounts can stay private: collect in full from your approved organizations only, and from personal or unknown accounts we collect metadata alone, their prompts and transcripts never leaving the machine. The browser extension hashes conversation text rather than storing it.
Where is our data stored, and how is it protected?
By default in AWS US data centers, unless your agreement says otherwise. It is encrypted in transit (TLS) and at rest with industry-standard encryption, kept for 60 days at most, and never used to train models, by us or our subprocessors. External penetration tests run at least once a year. More on our Security page.
Can we review your SOC 2 report?
Yes, on eligible plans: request access to our SOC 2 Type II report for your security review. Our compliance program also aligns with global data protection regulations, such as GDPR and CCPA.
How do I cancel my account?
Contact us, at any point. We won't keep anyone's hard-earned money if they're not satisfied with our services. After cancellation, you'll still have access to all the paid features until the end of your billing cycle.
Have more questions? Get in touch and we'll help.




