OpenAI Dots are always-on AI agents inside ChatGPT. Each one runs on its own cloud computer and keeps working after you close the app. They are rolling out now, but only to some paid plans, and not everywhere.
Here is who gets them, what they cost, and the one privacy detail worth knowing before you switch one on.
Quick facts:
- Announced September 29, 2026 at OpenAI DevDay, powered by GPT-6 Astra.
- One agent is included with ChatGPT Pro and Business Premium.
- Usage is free of plan limits for the first month, then plan-specific terms apply.
- Pro users in the EEA, Switzerland and the U.K. are not included yet.
- Extra agents and more capacity are planned, with no prices announced.
The first-month grace period matters most if you signed up at launch, because it is about to end.
What a Dot actually does day to day
You give a Dot a goal, and it keeps pushing the work forward without you approving every step. It can juggle several projects at once and learns your preferences from feedback.
When you have given it nothing to do, it looks for ways to help on its own. OpenAI calls this proactive research, and during it the agent’s app connections are read-only. That limit is sensible, since it stops an idle agent from sending or changing anything by accident.
You can message or call a Dot from ChatGPT on desktop, web or mobile, or from Slack and Teams. Context carries across all of them. With your permission, an agent can also work directly on your laptop, and it can reach more than 4,000 apps through OpenAI’s plugins. For another look at how far ChatGPT now reaches into your browser, see our post on the ChatGPT Chrome extension that controls your open tabs.
Who can use OpenAI Dots right now
Access depends on your plan and where you live. Some accounts may also wait a few days for the rollout to reach them.
| Plan | Access | Notes |
|---|---|---|
| ChatGPT Pro | Yes, one agent included | Not available in the EEA, Switzerland or the U.K. yet |
| Business Premium | Yes, one agent included | Rolling out in stages |
| Enterprise, Edu, Healthcare | Beta | A workspace admin must turn it on |
| Other plans | Not announced | No details from OpenAI so far |
Companies also get a preview of specialist agents. These have their own identity and credentials for one narrow job, such as invoice processing. Microsoft is working on managing them through its Agent 365 controls.
What OpenAI Dots costs after the free month
For now, the answer is “it depends.” OpenAI says agent usage will not count against your plan allowance for the first month. After that, terms are set per plan.
Extra agents, faster speeds and more monthly capacity are coming later, at prices OpenAI has not shared. If you run agents that work around the clock, watch your usage closely once the free period ends. Always-on tools can burn through allowances faster than a chat you start and stop.
The privacy question: HTTPS traffic and OpenAI’s certificate
This part is a user report, not an OpenAI statement. In a GitHub issue opened October 23, a developer says web requests from a Dots cloud computer go through a proxy that presents a certificate signed by OpenAI. In their case, that caused a connection error when calling a site by its raw IP address.
In plain terms, the agent’s secure web traffic passes through OpenAI’s own proxy before it reaches a site. That is a common setup for managed environments, and OpenAI has not publicly explained it. The issue had no reply from OpenAI when we checked.
So what should you do? Treat the cloud computer like a shared work machine. Do not let a Dot handle anything you would not want OpenAI’s systems to see.
OpenAI also says a monitoring system can pause or stop an agent, and that risky actions go through an “auto-review” check first. Those are useful safeguards, but they do not replace your own review. Our earlier report on OpenAI agents that disabled their own monitoring is a good reminder of why.
How to try a Dot safely
Start small and keep the stakes low for the first week. These steps help:
- Give it one narrow, low-risk job, such as summarizing customer feedback.
- Keep app connections read-only until you trust the results.
- Skip laptop access until you have seen how it behaves in the cloud.
- Read what it did at the end of each day, especially anything sent to other people.
The goal is to build trust gradually. An agent that works while you sleep is only useful if you can check its work in the morning.
Frequently Asked Questions
What is OpenAI Dots?
It is a set of always-on ChatGPT agents, each with its own cloud computer. You give them goals, and they work in the background and report back.
Is OpenAI Dots free?
One agent is included with ChatGPT Pro and Business Premium. Usage does not count against your allowance for the first month, and plan terms apply after that.
Can I use Dots in the EU or U.K.?
Not on Pro yet. Pro subscribers in the EEA, Switzerland and the U.K. are excluded from the current rollout.
Can a Dot control my computer?
Only with your permission. Without it, the agent works on its own cloud computer, which you can open at any time to watch.
Are my files and logins safe with Dots?
OpenAI says it uses auto-review checks and monitoring, and warns that agents can still make mistakes. Share only what the task needs.
My take: Dots are worth a cautious try if you already pay for Pro and have a boring, repeatable chore to hand off. Skip them for anything involving money, private files or customer data until OpenAI explains its proxy setup and shows what the paid terms look like. Use the free month to test, not to trust.
Leave a Reply