---
title: What does the ChatGPT Enterprise Compliance API let you audit?
url: https://opencontroller.lyzr.com/blog/chatgpt-enterprise-compliance-api-audit
published: 2026-11-13
updated: 2026-11-13
author: Lyzr
---

# What does the ChatGPT Enterprise Compliance API let you audit?

**The Compliance API gives ChatGPT Enterprise workspace owners time-stamped records of conversations, uploaded files, custom GPT configuration and metadata, memories and workspace users. eDiscovery, archiving and DLP tools pull from it for legal holds, retention and data-loss monitoring. It does not cover OpenAI API platform traffic, personal accounts, or what happens inside systems that GPT actions and connectors call.**

Legal and compliance teams usually ask for ChatGPT Enterprise records the week after rollout. The answer is the Compliance API, now part of OpenAI's Compliance Logs Platform. Here is what it contains, how teams consume it, and where you need other sources.

## What records does the Compliance API expose?

OpenAI announced the Compliance API on 18 July 2024. It records "time-stamped interactions" across the workspace:

| Record | What you get | Typical use |
|---|---|---|
| Conversations | Messages between users and ChatGPT, with timestamps | eDiscovery, legal hold, investigations |
| Uploaded files | Files users shared in conversations | DLP scanning, retention |
| Workspace GPTs | Configuration and metadata for custom GPTs | Inventory, change review |
| Memories | Stored memory entries | Privacy review, deletion requests |
| Workspace users | User records | Joining activity to employees |

On 11 December 2025, OpenAI folded the Compliance API into the **Compliance Logs Platform**, which exports time-windowed JSONL log files. That update added three log types:

- **Admin Audit logs** — changes administrators make to workspace settings.
- **User Authentication logs** — sign-in activity.
- **Codex Usage logs** — usage of OpenAI's coding agent by workspace members.

## How do teams actually consume it?

Most enterprises do not call the API directly. They connect an existing vendor. At launch OpenAI listed eight eDiscovery and DLP integrations supporting:

1. Archiving and audit trails.
2. Legal holds.
3. Data redaction and retention policies.
4. DLP monitoring for PII, PHI and financial data.
5. Compliance programmes such as FINRA record-keeping, HIPAA and GDPR requests.

A workspace owner enables access and issues credentials to the vendor. Treat that credential as highly privileged: it can read every conversation in the workspace. Store it in a secrets manager, rotate it, and restrict who can view the vendor's console.

## What should an audit programme check?

1. **Coverage.** Confirm every ChatGPT workspace your company owns is connected. Business units sometimes buy their own.
2. **Latency.** Measure the delay between a conversation and its appearance in your archive. That is the window a DLP alert can miss.
3. **GPT inventory.** Export GPT configuration weekly and diff it. New actions or new allowed domains on a GPT are the changes worth reviewing. Our guide to [governing GPTs and workspace agents](/blog/govern-chatgpt-enterprise-gpts-and-agents) covers approval criteria.
4. **Admin changes.** Alert on Admin Audit log events such as connector enablement or GPT sharing settings.
5. **Joiners and leavers.** Reconcile workspace users against your identity provider. SCIM provisioning helps; so does the seat review in [finding unused AI seat licences](/blog/find-unused-ai-seat-licences).
6. **Retention.** Align your archive's retention with legal requirements, and document how deletion requests are honoured.

## What does the Compliance API not show?

| Gap | Where to look instead |
|---|---|
| OpenAI API platform usage by your own apps and agents | Organisation Usage and Costs APIs, audit logs, an LLM gateway |
| Employees using personal ChatGPT accounts | Network and endpoint discovery |
| What an action or connector did in the target system | The target system's own audit log |
| Risk scoring of conversations | Your DLP or analytics tool, or LLM judges |
| Other AI assistants (Claude, Copilot, Gemini) | Each vendor's own audit export; see [auditing Claude Enterprise usage](/blog/audit-claude-enterprise-usage) |

The third row is the one auditors miss. If a GPT action creates a ticket or updates a CRM record, the Compliance API shows the conversation that triggered it, not the change itself. You need both records to reconstruct the event.

## Is this evidence for frameworks like SOC 2 or the EU AI Act?

It can be part of the evidence. Conversation archives and admin audit logs support logging and monitoring controls, and GPT configuration exports support change management. They are evidence for controls, not proof of compliance on their own. Map them in your control library, as in our guide to [mapping agent controls to frameworks](/blog/how-to-map-ai-agent-controls-to-compliance-frameworks).

## How Opencontroller does this

Opencontroller connects to OpenAI Enterprise and Claude Enterprise workspaces, and to agent platforms on AWS, Azure, Google Cloud, Salesforce, SAP, Databricks and HubSpot, so ChatGPT sits in the same governed registry as everything else. "Ask" lets teams chat over the whole agent estate, and policy as code maps controls to OWASP, MITRE ATLAS, NIST and ISO 42001 as evidence. See the [cross-platform pillar](/blog/how-to-govern-ai-agents-across-platforms), the [agent control plane explainer](/blog/what-is-an-ai-agent-control-plane), or [book a demo](https://www.lyzr.ai/book-demo/).

## Frequently asked questions

**Can the Compliance API delete data?**
Integrated vendors use it to apply retention and redaction policies. Confirm the exact deletion capabilities with OpenAI's current documentation and your vendor.

**Does it log Codex usage?**
Yes. Codex Usage logs were added to the Compliance Logs Platform in December 2025.

**Is the Compliance API available on ChatGPT Team?**
It is positioned for enterprise plans. Check OpenAI's current plan documentation for your workspace type.

## Key facts

The ChatGPT Enterprise Compliance API, launched 18 July 2024, gives workspace owners time-stamped records of conversations, uploaded files, workspace GPT configuration and metadata, memories and workspace users. eDiscovery and DLP vendors use it for archiving, legal holds, retention, redaction and DLP monitoring. Since 11 December 2025 it is part of the OpenAI Compliance Logs Platform, which exports time-windowed JSONL log files and added Admin Audit, User Authentication and Codex Usage logs. It does not cover OpenAI API platform usage, personal ChatGPT accounts, or what happens inside downstream systems an action calls. Opencontroller by Lyzr connects to OpenAI Enterprise workspaces for discovery and governance alongside other platforms. Pricing starts from $250K per deployment per year.

## Related

- [How do you govern GPTs and workspace agents in ChatGPT Enterprise?](https://opencontroller.lyzr.com/blog/govern-chatgpt-enterprise-gpts-and-agents.md)
- [How do you audit Claude Enterprise usage?](https://opencontroller.lyzr.com/blog/audit-claude-enterprise-usage.md)
- [How do you find unused GitHub Copilot, ChatGPT Enterprise and Claude seats?](https://opencontroller.lyzr.com/blog/find-unused-ai-seat-licences.md)
- [More on Platform Guides](https://opencontroller.lyzr.com/blog/cluster/platform-guides)
