OpenAI Adds New ChatGPT Enterprise Controls as Companies Watch AI Spending

OpenAI is adding new usage analytics and spend controls for ChatGPT Enterprise, a move aimed at companies that are watching AI costs more closely.

The update, announced on June 18, 2026, gives enterprise admins more visibility into how ChatGPT and Codex credits are being used across their organization. It also gives companies more control over limits, access and credit requests as AI becomes a bigger part of daily work.

The timing matters. Many businesses spent the last two years experimenting with generative AI. The early question was simple: what can this technology do? Now the question has changed. Companies want to know what AI is costing, who is using it and whether that spending is creating real value.

What the New Controls Do

The new controls are designed to help companies manage ChatGPT Enterprise usage in a more structured way.

Admins can view credit usage across ChatGPT and Codex in one place, with breakdowns by users, products and models. They can also set default usage limits for a workspace, create limits for specific groups and approve higher limits for individual employees who need more capacity.

Employees can see their own credit usage and request additional credits when needed. That gives companies a clearer approval path instead of simply blocking productive users or allowing unlimited spending without oversight.

This matters because different teams use AI in different ways. An engineering team using Codex every day may need more credits than a department that only uses ChatGPT for occasional summaries or research. With more flexible controls, companies can support heavy users while keeping spending visible.

Why Companies Care Now

AI budgets had a grace period while businesses were still testing what generative AI could do. That phase is starting to fade.

Finance and IT teams now want the same thing they expect from other major software tools: predictability, accountability and a clear view of usage. A company cannot judge the value of AI if it cannot see where the spending is going.

The controls also reflect a bigger shift in enterprise AI. ChatGPT is no longer only an experimental tool used by a few curious employees. In many companies, it is becoming part of coding, research, document analysis, customer support drafts and internal operations.

As AI becomes more useful, it also becomes harder to manage casually.

Visibility Is Not the Same as Value

Better controls can help companies avoid waste. They can show which teams are using AI heavily, where credit usage is rising and whether some groups need more training.

But visibility alone does not prove value.

A dashboard can show how much a company is spending. It cannot fully answer whether AI is helping employees do better work, save meaningful time or produce stronger results. That question still requires managers to look beyond the bill and study the actual impact on teams.

This is where companies need to be careful. Cutting costs too aggressively could slow down employees who are using AI well. Allowing usage to grow without review could create another expensive software problem.

The best approach is a middle path: clear limits, useful data and a simple way for employees to request more access when there is a real business need.

The Risk of Too Much Control

There is also a human side to this update.

Usage analytics can help companies manage budgets, but they can also make employees feel watched if handled poorly. The difference depends on how businesses use the data.

If companies use these controls to understand costs and support better AI adoption, the tools can be helpful. If they turn them into another layer of worker monitoring, trust could suffer.

That is not really a software problem. It is a management problem.

Why This Feels Like a Turning Point

OpenAI’s new ChatGPT Enterprise controls show that workplace AI is entering a more serious phase.

The excitement around generative AI is still there, but businesses are now asking more practical questions. Who is using it? What is it costing? Which teams need more access? Where is AI actually helping?

Those questions may not sound as exciting as a new model launch, but they matter more for companies trying to use AI at scale.

Personally, I think this is a healthy shift. AI should save time, not quietly become another uncontrolled expense. But the real goal should not be only to reduce the bill. The better question is whether employees are getting more room to do meaningful work instead of being buried under repetitive tasks.

That is the number no dashboard can fully capture, but it may be the one companies should care about most.