OpenAI dots: what they do, and who can get one (not Pro users in Europe)
OpenAI's dots are AI helpers that work day and night on their own computer. What they do, what they cost, how you stay in control, and who has to wait.
On 29 September 2026, OpenAI launched dots. A dot is an AI helper that never switches off. It has its own computer in the cloud and keeps working while you sleep. But if you live in Europe and pay for ChatGPT Pro, you cannot get one yet. Only business plans can use dots in Europe for now. We read OpenAI's announcement and six of its help pages to explain what dots do, who can use them, and how to stay in control.
What is a dot?
A dot is a helper you give ongoing jobs to. Not one question, but a task that runs for days or weeks. For example: keep a list of everything for a team outing up to date, or watch customer feedback and suggest fixes.
OpenAI's documentation calls it "an always-on agent that keeps work moving and asks for your input when a decision needs you". An agent is an AI that works on its own, step by step.
A dot runs on GPT-6 Astra, OpenAI's most capable model. That is the same model we tested in our comparison of GPT-6.1 Sol and GPT-6 Astra. You can give your dot a name and pick how it looks: its shape, colour, eyes and even glasses.
What can a dot do?
A dot has its own computer and web browser in the cloud. That is why it can keep working when your own computer is off. According to OpenAI, it can do research, look at data, prepare documents and build software.
You can reach the same dot in several places:
- the ChatGPT app on your computer or phone
- Slack or Microsoft Teams, the chat tools many companies use
- a voice call, by pressing the phone button
It remembers your work. It keeps its own notes about your choices and plans, so you do not have to explain everything again.
It can also use your apps, such as email, files or code, through plugins. Plugins are small add-ons that connect ChatGPT to other services. OpenAI says there are more than 4,000. If you allow it, a dot can even use your own computer. But only while that computer is on and the ChatGPT app is open, OpenAI explains.
When a website asks for a password, the dot sends you a private sign-in form. You type your password there, not in the chat.
Who can use it? Not Pro users in Europe
Dots are coming to people step by step. OpenAI's help page lists who can get one:
| Plan | Who can use dots |
|---|---|
| Pro (100, 200 or 500) | People aged 18 or older, outside the European Economic Area, the United Kingdom and Switzerland |
| Business Premium | Rolling out worldwide |
| Enterprise | Rolling out worldwide, but only after the company's administrator switches it on |
The European Economic Area means the EU plus Norway, Iceland and Liechtenstein. So if you live in the Netherlands, Germany or France and pay for ChatGPT Pro yourself, you cannot get a dot yet. A company in the same country with Business Premium can, once the rollout reaches it.
OpenAI does not say why Pro users in Europe have to wait. We did not find a reason on its pages, so we will not guess. Even on an eligible plan, OpenAI warns, "you may not see dots immediately".
What does a dot cost?
Your first dot is free, as part of the plan you already pay for. OpenAI's announcement says: "Your first dot is included in your Pro or Business Premium plan at no extra cost."
There are a few details:
- Chatting is free of limits. Talking with your dot does not count toward your ChatGPT usage limits.
- Bigger jobs do count. When your dot starts tasks in Codex or ChatGPT Work, those count toward the limits as usual.
- Extra room at the start. Your plan includes an allowance for deeper work, with higher limits in the first month after launch.
- More later. OpenAI says you will be able to add more dots in the future, or make a dot faster or able to do more work per month. It does not say yet what that will cost.
How do you stay in control?
A helper that works on its own needs clear limits. OpenAI built in several.
It checks before big steps. Before a dot does something that affects your accounts or shares information, an automatic check looks at your instructions, your permissions and OpenAI's safety rules. Then the dot goes ahead, asks you first, or leaves the step to you. Some things are always yours to do, like changing a password, OpenAI says.
You can set your own rules. For example: "ask me before sending anything to customers" or "let me delete shared files myself". There are four choices: act without asking, act only when you say so, ask first, or hand the step to you.
Asking for a draft is not permission to send. OpenAI gives this exact example: if you ask your dot to write replies, that does not mean it may send them.
Stopping is not undoing. Pausing your dot stops its main task, but not every task it handed out. And anything it already did, like a sent message, stays done. OpenAI also says plainly: "Dots can still make mistakes, so always review consequential work."
What happens to your data?
A dot sees a lot: your conversations, the apps you connect and its own notes about you. So what does OpenAI do with that?
According to OpenAI's help page, your normal ChatGPT data settings also apply to your dot. OpenAI says it does not train its models directly on your dot's own research or private notes. If something from that research ends up in a normal conversation or task, your usual data settings apply.
Your dot also keeps notes across all the places you talk to it, OpenAI explains. Something you tell it in Slack can help it in ChatGPT. Before it shares something from a private conversation with other people, it checks that you allowed it.
If you delete your dot, that cannot be undone. And deleting it does not take back messages it already sent, OpenAI warns.
Should you try one?
If you can, it is worth trying carefully. A dot can take real work off your hands. But it also gets access to your accounts, your apps and maybe your computer. These five tips come straight from what OpenAI's own documentation recommends or warns about:
- Start small. Pick one task you check anyway, such as a weekly overview.
- Connect only what it needs. You can let it read your email without letting it send email.
- Set rules for big steps. For example "ask me before sending".
- Check its work. OpenAI itself says dots can make mistakes.
- Know how to stop it. Pausing does not stop every task, and it does not undo anything.
If you are a Pro user in Europe, you will have to wait. If your company has Business Premium, ask whoever manages it.
Giving an AI more power always needs clear limits. We saw that in our test of Claude Code mods, where home-made safety rules did not catch everything.
How we checked
On 6 October 2026 we read OpenAI's announcement of 29 September and six pages of its dots documentation: the overview, getting started, messaging, tasks and memory, computers and apps and controls. The screenshots are our own captures of those pages, cropped only. OpenAI's website showed us the Dutch version of the announcement first; we quoted the English version.
We could not try a dot ourselves. The owner of this site has a ChatGPT Pro plan in the Netherlands, and OpenAI's rules exclude Pro users there. So everything about how dots work is OpenAI's own description, not our test. The charts and the checklist were made by us from OpenAI's pages. None of the images was made by AI.
OpenAI says dots are rolling out step by step, so who can use them may change. We will update this article when dots come to Pro users in Europe.