# OpenAI dots in Europe: who can turn them on and what to set first > Dots launched on 29 September 2026, but according to the OpenAI Help Center a company in the European Economic Area gets them through Business Premium, not through a Pro account. We describe availability, four rule behaviours and five settings a company fixes before anyone turns an agent on. - **Canonical URL:** https://appwave.dev/en/blog/openai-dots-in-europe-availability-and-first-settings - **Language:** en - **Updated:** 2026-10-06 --- - **Published:** 2026-10-06 - **Author:** Mateusz Cieśliński - Co-Founder, DevOps Engineer - **LinkedIn:** https://www.linkedin.com/in/mateusz-cie%C5%9Bli%C5%84ski/ - **Reading time (min):** 6 - **Topics:** OpenAI dots, AI agent, AI security ![Four identical matte black spheres in a row, each with a short rail in front and a sliding block set at a different position. Only the last rail glows emerald, with a sine wave running behind.](https://cms.appwave.dev/uploads/openai_dots_in_europe_availability_and_first_settings_fbdf7e22a0.webp) **OpenAI dots, the background agents in ChatGPT, launched on 29 September 2026, but according to the OpenAI Help Center, in the European Economic Area a company gets them through a Business Premium plan or an Enterprise workspace admin, not through a Pro account. Before anyone turns a dot on, set five things: who creates it and which apps it connects, which actions it takes alone, whether it can use a computer, what happens to its memory when an app is disconnected, and who reviews its work every week. Disconnecting an app does not delete what the dot already pulled from it.** Availability is the first question, and the settings are the second, more expensive one. A dot is not a chat where a mistake shows at once. It works while you are not looking, so what you set before launch will be the only control you have at night and at weekends. ## What we see in the market: dots availability depends on the plan OpenAI describes a dot as an agent that has its own cloud computer, works across connected apps and remembers context between conversations. The GPT-6 Astra model runs it. The announcement of 29 September 2026 speaks generally about eligible markets. The details are in the Help Center, and that page decides who gets a dot in the EEA. - **Pro.** Available in the EEA: no. Notes: rolling out in markets excluding the EEA, Switzerland and the UK. - **Business Premium.** Available in the EEA: yes, according to the Help Center. Notes: all supported ChatGPT regions. The EEA and Poland are not named there. - **Enterprise, Edu, Healthcare.** Available in the EEA: beta once an administrator enables it. Notes: off by default at launch. ![Decision map of dots availability in the EEA. The question about the company's ChatGPT plan branches into three plans: Pro with no access in the EEA, Business Premium available according to the OpenAI Help Center, and Enterprise, Edu and Healthcare in a beta enabled by an administrator.](https://cms.appwave.dev/uploads/openai_dots_in_europe_availability_and_first_settings_diagram_5b98b22423.webp) The rollout is gradual, so even with an eligible plan access may take a few days to appear. Your first dot costs nothing on top of a Pro or Business Premium plan. We do not give plan prices: **Data not available**, we did not check OpenAI’s price list for Poland. ## Four rule behaviours: which actions to assign to them A dot starts with built-in rules for when to act alone and when to ask. In Custom rules you describe an action and then pick one of four behaviours. According to OpenAI, “pre-approved” means that you requested the action yourself in the prompt. The split below is our proposal for a company of 10-200 people, not a recommendation from the vendor. - **Take action without asking.** Example actions: reading and summarising email, drafting documents inside the dot’s own area. Why it sits here: the dot sends or changes nothing outside, and it is easy to undo. - **Take action if pre-approved.** Example actions: sending an email you explicitly asked for in the prompt. Why it sits here: a person made the decision, the dot only carries it out. - **Ask before taking action.** Example actions: changes in systems, anything involving customer data. Why it sits here: a third party sees the effect, or it is hard to reverse. - **Hand off to you.** Example actions: payments, contracts, personal data. Why it sits here: responsibility does not move to the tool. According to the announcement, OpenAI also keeps a few things outside the rules. Changing a password always stays with the person, and built-in safety requirements apply whatever you set. On top of that there is an automatic review of actions that could affect your accounts or share information, which checks them against your instructions, rules and safety requirements. We describe the same split into modes at more length in our text on [human in the loop](https://appwave.dev/en/blog/human-in-the-loop-ai-human-oversight-in-a-small-business). ## Five settings before the first run **Who creates the dot and which apps it connects.** You pick the apps yourself and manage their permissions in ChatGPT settings. Check the permission scope of each integration separately. When you are not talking to it, a dot looks for ways to help using connected apps through read-only tools, which according to OpenAI cannot send messages, change content in apps or control a browser or computer. That is a good boundary, but the dot is then reading your data although you asked for nothing. **Rules written down before launch.** The table above is a starting point. Every action on the list gets one of the four behaviours, and the ones you cannot assign land on “Ask before taking action.” **Local computer: leave it off.** Access is optional and starts turned off. Once on, according to the Help Center a dot can create Work or Codex tasks on that computer, use local skills, and use the local browser when its own is blocked. Switch it on for a specific role, not for everyone. **Disconnecting is not deleting.** This is the most common surprise. Disconnecting an app does not delete the information the dot has already obtained, and the memories it built from it can only be removed by resetting the dot, together with its conversations and scheduled tasks. Before you connect a mailbox, decide whose mailbox it is. **An owner of the review.** One named person, once a week, in the dot’s activity view. We assume 30 minutes a week per dot \[estimate: AppWave assumption, not a measurement\]. Without an owner, the rules from the previous section are only a document. In our text on [hallucinations and AI agents](https://appwave.dev/en/blog/ai-agent-tools-three-things-nobody-mentions) we put it this way: “For anyone to catch that, logs have to exist, somebody has to read them, and reading them has to be in someone’s job description.” ## What dots do not do on launch day According to the Help Center (as of 5 October 2026), a dot is created from the ChatGPT desktop app or from ChatGPT in a desktop browser. Creating one on a phone is not possible for now, and the mobile browser version is not supported. A dot cannot have its own standalone email address at launch, so it uses a personal email account you connect. It cannot initiate calls to you. You can connect it to Slack, and texting is a limited beta for Pro users in the US, not available in Business or Enterprise workspaces. OpenAI announces “specialist dots,” agents with their own identity and access to company systems, as pilots for selected enterprises, integrated with Microsoft Agent 365. That is not an offer for a company of 30 people. ## What this means for your company **Probable conclusion.** For a company in the EEA, the realistic route to dots today is Business Premium or an Enterprise beta enabled by an administrator. A Pro account in the EEA does not give access. We base this on OpenAI’s documentation, not on our own test on an EEA account. **Probable conclusion.** If a dot is to write to customers or business partners, the transparency duties of the EU AI Act may come into play. We describe them in [Chatbot and the EU AI Act: what to check before 28 October 2026](https://appwave.dev/en/blog/chatbot-ai-act-checklist-before-28-october-2026). We are not a law firm, so a lawyer decides how a specific case is classified. **Hypothesis to test.** A dot reads like a personal tool: one person, their mail, their calendar. We do not yet know how it holds up in a team where three people want different things from one dot. We will come back to this once we see it at a client. Why a general-purpose agent alone rarely fits a company’s own process, we wrote in [An AI agent alone is just a conversationalist. Skills make it work](https://appwave.dev/en/blog/agentic-ai-skills-why-an-agent-alone-is-not-enough). The cost of waiting runs the other way than usual. Waiting costs little, because dots are new and OpenAI is rolling them out gradually. Haste costs, if someone connects a mailbox before the rules are set. ## What happens when you contact us Most AI agencies deliver a system. We deliver understanding, so we start with the question of whether you need an agent at all, not with configuring a tool. We run a [two-week AI audit](https://appwave.dev/en/services/ai-audit) with scope and price agreed before we start. It maps processes, estimates the return and ends with a report even when the conclusion is “don’t deploy this.” Deployment is a separate decision, and if you want an agent built around your own process, it is covered by our [AI agents service](https://appwave.dev/en/services/ai-agents). We build with you, not for you. The rules for an agent, meaning what it does alone and what it hands to a person, we write together with the person who answers for it. The first call lasts 45 minutes, and a team subscription starts at 3,875 PLN net a month. How to judge for yourself whether a process suits an agent at all, we describe in [Not every company needs AI. Let’s find out whether yours does](https://appwave.dev/en/blog/does-your-company-actually-need-ai). This text describes OpenAI’s documentation of 29 September 2026, not our deployments. We are not an OpenAI partner, we do not sell ChatGPT licences, and we do not promise that the settings will work for you exactly as described. [Book a 45-minute call](https://appwave.dev/en/appointment). ## Sources All pages accessed on 5 October 2026. - OpenAI, 29 September 2026: [Introducing dots](https://openai.com/index/introducing-dots/), sentences on passwords, the automatic review, read-only tools and “specialist dots” pilots. - OpenAI, Help Center, page checked on 5 October 2026: [Getting started with your dot](https://help.openai.com/articles/20001530), sections on plans, computer, apps, rules and reset. - AppWave, 20 August 2026: [Hallucinations are the smallest of three problems with AI agents](https://appwave.dev/en/blog/ai-agent-tools-three-things-nobody-mentions). - AppWave, service: [AI audit](https://appwave.dev/en/services/ai-audit). ## Questions & Answers (FAQ) ### Are OpenAI dots available in Europe? It depends on the plan. According to the OpenAI Help Center (as of 5 October 2026), dots are rolling out to Pro users in markets excluding the European Economic Area, Switzerland and the UK, so a Pro account in the EEA will not get them. Business Premium covers all supported regions, and Enterprise has a beta that a workspace admin enables. ### What are OpenAI dots? They are background agents in ChatGPT, built on the GPT-6 Astra model. Each dot has its own cloud computer, works across connected apps and remembers context between conversations. OpenAI announced them on 29 September 2026 and warns that a dot can make mistakes, so important results need checking. ### Can a dot use my computer? Only if you allow it. Access to a local computer is optional and starts turned off, and you switch it on in the ChatGPT app on that computer. By default a dot works on its own cloud computer, separate from your files. ### Does disconnecting an app delete the data a dot already pulled? No. According to the OpenAI Help Center, disconnecting an app does not delete information the dot has already obtained from it. To remove it you have to delete the dot itself with Reset, which also deletes its conversations, saved memories and scheduled tasks. ### How do I set what a dot does alone and what needs approval? In Custom rules you describe an action and pick one of four behaviours: take action without asking, take action if pre-approved, ask before taking action, or hand off to you. Actions involving money, personal data and contracts are better left on the last two. --- ## About AppWave AI development company from Łódź, Poland. We build custom AI software, automation agents and web applications. We co-create them with the client and guarantee every system we deliver. - **Legal name:** AppWave sp. z o.o. - **Address:** ul. Kolumny 147E/1, 93-611 Łódź, Poland - **Phone:** +48 538 441 413 - **Email:** office@appwave.dev - **Business hours:** Monday-Friday, 09:00-17:00 - [Free consultation](https://appwave.dev/en/appointment)