OpenAI ChatGPT Enterprise governance guidance

Reviewed 3 October 2026. Aicura is not affiliated with OpenAI.

About OpenAI ChatGPT Enterprise

OpenAI ChatGPT Enterprise gives staff a general purpose AI assistant for drafting, analysis, research and working with their own files, inside a workspace the organisation administers. Its reach grows as the workspace is extended. Custom GPTs package instructions and knowledge for a team, connected apps let it read and act in other business systems as the user and agent capabilities can complete multi-step tasks with a browser, code and those apps. Good governance lets an organisation open up each of these with confidence and keep oversight in step as the workspace grows.

Where governance makes the difference

• Information going further than intended. Staff can paste or upload anything into a general purpose assistant, and custom GPTs can be shared across the whole workspace. Without clear rules and sharing limits, sensitive information can end up in places and with people it was never meant for.
• Connected apps acting with the user’s full reach. Apps act with the permissions of the person using them. Where write actions are switched on broadly, an assistant can change records, send messages or move files across systems well beyond what anyone intended to allow.
• Content steering the assistant. An AI assistant follows instructions in what it reads, so a web page, document or email it is working with can influence what it does with internal information and connected systems.

---

For the executive team

A general purpose assistant needs clear rules of use

OpenAI ChatGPT Enterprise can help with almost any task, which is its value and the reason it needs plain rules on what information may go into it and what its output may be used for. Those rules work best when staff understand them and the approved workspace is the easy choice over personal AI tools.

Every extension needs an owner

Custom GPTs, connected apps and agent capabilities each extend what the assistant can reach and do. Each one needs a named person accountable for it, a clear purpose and agreed limits, with the most capable extensions opened up deliberately rather than all at once.

Oversight needs to keep pace with change

New AI versions, new apps and new capabilities arrive regularly, and older AI versions are retired. The workspace approved at launch will not be the one running six months later, so oversight works best as an ongoing practice rather than a one-off sign-off.

Questions worth putting to management

• Who is accountable for OpenAI ChatGPT Enterprise, and who owns each custom GPT and connected app?
• What information are staff permitted to put into it, and how is that communicated?
• Which business systems can it act in, and who decided that?
• How much AI use is happening outside the approved workspace, and what is being done about it?
• How would the organisation know if information was shared too widely or an action went beyond what was intended?
• If a regulator, customer or insurer asked how the assistant is governed, what would be shown to them?

What good reporting looks like

Expect regular reporting on adoption across the organisation, which extensions and connected systems are in use, the main risks and how they are changing, AI use outside the approved workspace, any incidents or near misses and the significant decisions taken, with who made them and why.

How Aicura can help

• One place showing OpenAI ChatGPT Enterprise, every custom GPT and connected app built on it, who owns each one and what it can reach.
• Board reporting drafted from the governance record, covering adoption, risks, incidents and the decisions taken.
• A public trust centre where customers, partners and regulators can see how the organisation governs its AI, backed by evidence a third party can verify.

---

For risk, compliance and assurance

Set where OpenAI ChatGPT Enterprise sits in the risk appetite

Decide which classes of information may be entered or uploaded and which decisions its output may inform. Output that feeds decisions about people, such as hiring, performance or customer outcomes, needs review by someone who knows the subject before it is relied on. Content prepared for customers or the public needs the same review, given the reputational stakes.

Assess the workspace and each extension separately

Assess the workspace as a whole before rollout, covering who has access, what may be entered and how sharing is set. Assess each custom GPT, connected app and agent capability on its own terms, covering the data it reaches, whether it can act or only read, which external services it calls, whether its output affects people and who reviews it.

Reassess when anything changes

A new connected app, a move from read-only to write actions, a newly enabled capability or a change to the underlying AI all move the workspace away from what was assessed. Each needs to trigger a fresh assessment rather than wait for the annual cycle.

Make the data decisions explicitly

Decide how long conversations and files are kept, who may access activity records and for what purpose, and let staff know. Decide where data needs to be stored and processed, then confirm which content and capabilities the workspace’s residency arrangements actually cover. Decide whether web browsing and agent capabilities are appropriate for teams handling the most sensitive work.

Confirm which protections apply

Check which accounts and workspaces carry the vendor’s business data commitments, including how business data is used for training and how it is retained. Work done in consumer AI tools on personal accounts sits outside those arrangements, so find that use and bring it into the approved workspace.

Set limits and name who acts

Agree the limits that matter, such as unusual volumes of write actions through connected apps, sensitive information appearing in prompts, GPTs shared beyond their intended audience or a sudden change in use, and name the person who acts when one is crossed.

Be ready to show governance to others

• Policy. Maintain an AI use policy that covers OpenAI ChatGPT Enterprise specifically, including what may be entered into it, who may build and share GPTs, which apps may act and how output affecting people is reviewed. Tie each commitment to the control that carries it out.
• Accountability. Record committee and management decisions with their reasons when they are made. Have extension owners confirm periodically that their GPTs and apps still match what was approved.
• Operating evidence. Keep records that controls actually ran, such as access reviews, extension approvals, reviewed outputs and interventions.
• External scrutiny. Prepare for internal and external audit, certification against standards such as ISO/IEC 42001, regulator enquiries and insurer questions about AI controls.
• Customers and partners. Due diligence questionnaires increasingly ask whether customer information goes into AI assistants. Have the answer ready and supported by evidence.
• Incidents. Be able to reconstruct which user, which conversation, which app or GPT and which controls were in place at the time, and keep activity records under the organisation’s own retention policy rather than relying on the vendor’s.

How Aicura can help

• Risk and impact assessments against recognised frameworks for the workspace and each extension, run again whenever something material changes and feeding a managed set of risks and controls that can mirror to the GRC platform.
• Reviews of the AI use policy against recognised frameworks, with committee decisions and owner attestations recorded against the systems they concern.
• Limits on activity with responses the organisation chooses, from alerting the owner to re-running an assessment or opening an incident.
• Every decision and control action kept as evidence that can be verified without relying on the organisation’s word.

---

For technology and security

Discover every form OpenAI ChatGPT Enterprise takes

Inventory the workspace, every custom GPT, every connected app and custom connector and each agent capability that is switched on. Record the external domains GPT actions call and the business systems each app reaches. Include the desktop app on managed devices and any AI use on personal accounts in the same picture.

Build access on roles and groups

Use role-based access with groups synced from the identity provider, so access to GPTs, apps, capabilities and AI versions follows a person’s role rather than individual settings. Keep owner and admin roles to a small number of people and review membership on a schedule.

Treat connected apps as access to the systems behind them

Apps act with the permissions of the signed-in user, so tidy access in the connected systems before enabling them. Start with read-only actions, open write actions only to the roles that need them and test with limited accounts before widening. Restrict GPT actions to approved external domains and keep that list short.

Put change points under change control

Enabling an app, switching on write actions, publishing a GPT to the workspace, enabling a new AI version or switching on agent capabilities all change what the workspace can do. Route each through a request and approval, and run a preflight check against agreed rules before the change goes live, so the answer is recorded alongside the decision.

Govern the agent capabilities deliberately

Agent capabilities can browse, run code and act through apps across many steps. Set approval requirements, web access and network allow lists centrally, enable them first for teams whose app access has already been reviewed and test both permitted and blocked actions before widening.

Use the network for what the request path cannot reach

The assistant’s processing runs in the vendor’s cloud through its own application, so an AI gateway in the request path cannot inspect conversations. Manage traffic from managed devices at the network edge instead, allowing the approved workspace sign-in and steering people away from personal accounts and AI services outside it.

Feed activity records into monitoring

The workspace’s compliance and audit records cover conversations, files, GPTs, sign-in events and agent activity. Export them into monitoring and the SIEM through a thin transform, apply the limits risk has agreed and set responses ranging from alerting the owner to removing a GPT or withdrawing an app. Keep the exported records under the organisation’s own retention and legal hold processes.

Watch for changes to the underlying AI

New AI versions arrive in the workspace and older ones are retired on notice. Track which versions each team uses, decide before enabling a new one and update defaults and saved settings ahead of a retirement so workflows built on the assistant keep behaving as expected.

How Aicura can help

• One inventory of OpenAI ChatGPT Enterprise, its GPTs, connected apps and the AI versions behind them, with discovery of AI use on managed devices that is not yet recorded.
• Preflight checks before an app or capability is enabled, a GPT goes live or a new AI version is used, with every answer recorded.
• A collector inside the network that takes in the workspace’s compliance and audit records, holds them against agreed limits and streams events to the SIEM.
• Notice to every affected entry when the vendor changes or retires an AI version, with the organisation’s position and response recorded.

---

Getting lasting value from OpenAI ChatGPT Enterprise

OpenAI ChatGPT Enterprise pays off when it is governed as an ongoing practice rather than approved once at launch. Three things matter whatever the role. The first is knowing everything that is in use, including every GPT and connected app, and who is accountable for each. The second is keeping access, controls and assessments in step with a workspace that keeps gaining reach. The third is being able to show, at any time, how decisions were made and that controls held. Organisations that get these right can extend the assistant further and faster, with confidence rather than caution.

---

Sources

• OpenAI, Compliance API and audit events
• OpenAI, Governance
• OpenAI, Plugin controls
• OpenAI, GPTs and sharing
• OpenAI, Roles and workspace permissions
• OpenAI, Workspace model availability
• OpenAI, ChatGPT Work overview
• OpenAI, Agent security
• OpenAI, ChatGPT Work cloud security