The dots admin preview for ChatGPT Enterprise, Edu and Healthcare workspaces ends November 18, 2026. On that date, dots become available to members by default unless an admin opts out. Billing also begins for every enabled dot.
Dots are OpenAI’s always-on agents in ChatGPT. Each one has its own cloud computer, works through the apps a member connects, and keeps working between conversations. We explained the product in What Are ChatGPT Dots? and covered personal setup in How to Set Up a ChatGPT Dot. This post is for the people who run the workspace.
The short version: the dots admin preview leaves you 12 days. Decide whether to opt out or enable dots for chosen roles. Check the Agent Security settings dots need. Look at what connected apps expose to proactive research. Then get pricing in writing from your OpenAI rep, since OpenAI has not published Enterprise dollar rates as of November 6, 2026.
What OpenAI changed for Enterprise and Edu
OpenAI’s Getting started with your dot help article says dots "are generally available to eligible ChatGPT Enterprise, Edu, and Healthcare workspaces." It sets the dots admin preview from November 4 through November 18, 2026. During it, members can use dots only if an admin enables access for their role.
Then the default flips. Per the same article, starting November 18 dots "become available by default unless your admin opts out."
Three more limits from that article matter for planning:
- Dots are not available for workspaces with UAE data or inference residency.
- Each member can create one personal dot.
- Healthcare and regulated workspaces should review OpenAI’s HIPAA eligibility list first.
OpenAI’s Introducing dots post from September 29 still calls Enterprise access a beta that admins can enable. The help article is newer, so we follow it here. A November 3 update on that post says dots can now draw on GPT-6.1 Terra as well as GPT-6 Astra. We covered that model in our GPT-6.1 Terra post.
What the dots admin preview means for your workspace
During the dots admin preview, dots stay off unless you turn them on. OpenAI’s Managing dots in ChatGPT workspaces article says "dots are off by default" while the preview runs.
Two points in that article change how much risk the default carries. First, enabling access "does not automatically create dots for members." Each member creates their own dot through onboarding. Second, billing applies to enabled dots. So November 18 gives every member permission to create a dot. It does not create a fleet of agents overnight.
Once a member creates a dot, it can connect to their apps, research in the background and start billed work. If you have no agent guidance for staff yet, November 18 decides for you.
OpenAI also tells admins to review effective workspace and role permissions "before and after that transition." Take that literally. One role that grants dots can give a member access even when another role turns it off.
Where the dots controls live
The path is Workspace settings, then Permissions & roles, then Permissions, then Workspace default. Open Workspace capabilities and find Use dots. Custom roles can give specific groups different access.
| Setting | What it controls | Default, per OpenAI |
|---|---|---|
| Use dots | Whether members can use dots | Off during preview; available from November 18 unless you opt out |
| Add dots to Slack and Microsoft Teams | Lets a dot join Slack or Teams and post as itself | Not stated; Teams has limited availability |
| Allow local computer access | Lets a dot use a member’s local files and run commands | Off by default for Enterprise |
| Use custom rules for dots | Lets members write rules for their dot’s actions | Off by default for Enterprise |
| Cloud browser use, cloud network access, cloud computer use | What a dot can do on its own cloud computer | Existing browser and network settings carry over |
| Use password manager | Password manager use with dots and Work Cloud | Not stated |
The cloud controls apply to dots even if ChatGPT Work or Work Cloud is off. Website allow and block lists in Agent Security apply to a dot’s cloud browser. Support for OpenAI’s broader network policies on dot cloud tasks "is still in development," per the article. Plan around that gap.
One more line to note. The article says Enterprise model controls and default model settings "do not apply to dots." If you restrict models elsewhere in ChatGPT, that limit does not reach dots.
Agent Security requirements come first
Dots need compatible Agent Security policies, according to the Managing dots article. Your security admin should open Admin Console, then Agent Security, then Global, then Requirements.
Two values must be allowed:
- Allowed approval policies must include on-request.
- Allowed approval reviewers must include auto-review.
OpenAI asks admins to keep their other allowed values and defaults. It also says to check other saved policies for conflicts before saving. If your policy is tight for ChatGPT Work or Codex, test the change on a pilot role.
Auto-review is OpenAI’s own checker. The launch post says it checks actions that could affect accounts or share information. It weighs them against instructions, Custom Rules and safety requirements. Our ChatGPT Enterprise admin checklist covers the earlier Work and Codex settings this builds on.
Proactive research and connected apps
The Managing dots article says there is "no separate admin setting to turn off proactive research" for dots. The Getting started article repeats this from the member side.
Proactive research is what a dot does when no one is talking to it. Per the launch post, it uses apps a member already connected, through read-only tools. The Getting started article adds that a dot can form memories from connected information on its own. Disconnecting an app later does not delete what the dot already learned.
So your real control is app access. Enabling dots "does not grant access to every app or website," per the Managing dots article. Plugin controls decide which apps are available and what they can do. App permissions decide when an action needs approval. Each service’s own authorization sets the outer limit.
Before the dots admin preview ends, list the apps your members can connect to ChatGPT. Mark the ones that hold donor records, student data, member files or HR data. Those are the apps a dot can read in the background.
Slack, Teams and texting
For Slack, a member needs both Use dots and Add dots to Slack and Microsoft Teams. If your Slack workspace requires app approval, a Slack owner or app manager must approve it too.
Only a dot’s owner can direct it. But everyone in a channel can see the messages a dot posts there. Set expectations with staff before dots show up in shared channels.
Texting is simpler. Per the Managing dots article, phone messaging through iMessage, RCS or WhatsApp is unavailable for Enterprise.
What dots cost on Enterprise
Pricing is the thinnest part of what OpenAI has published. Enterprise workspaces can pay "through flat rates mapped to the type of work you want your dot to achieve, or scaled consumption," per the Getting started article. It gives no dollar figures, as of November 6, 2026.
The dates are clear. Dots are free during the dots admin preview, November 4 through 18. From November 18, billing begins for all enabled dots. Work a dot hands to ChatGPT Work or Codex threads is billed separately at standard rates.
Admins have some cost levers. Per the Managing dots article, an admin pause stops billing for that dot. Removing a member’s access does too. Workspace analytics shows enabled-dot counts, and each dot’s details show its saved cost tier.
Pro and Business Premium are different. The launch post says the first dot comes with those plans "at no extra cost," with extended limits for the first month. OpenAI has not said what those limits become after that.
Data use and offboarding
On training, the launch post says OpenAI does not use content from Business, Enterprise or Edu workspaces to improve its models by default. It also says OpenAI does not train directly on proactive research or a dot’s notes to itself.
Offboarding needs a new step. Removing or deprovisioning a member revokes access but "does not automatically delete their dot," per the Managing dots article. Separate Work, Codex or scheduled tasks may keep running after a pause. Add dots to your offboarding list: apps, website sign-ins, local computers, Slack identity and delegated tasks.
Owners can review every personal dot in the workspace, with its creator, status and cost tier. Admin pause, resume and delete actions appear in audit logs.
Your checklist before the dots admin preview ends
-
Decide the default. Opt out, or keep the default and limit dots to chosen roles. Check every role, since one grant can override another.
-
Fix Agent Security. Allow on-request approval and the auto-review reviewer. Check saved policies for conflicts.
-
Review connected apps. Proactive research cannot be turned off separately. App access is your control.
-
Set the cloud controls. Decide on cloud browser, network, computer use and password manager. Leave local computer access off without a clear need.
-
Decide on Slack and Teams. Turn that permission on only for roles that need it.
-
Ask your OpenAI rep about cost. Get flat and scaled rates and the billing unit. Ask how Work and Codex handoffs appear on the invoice, and whether a spending cap exists.
-
Brief staff. Say who can create a dot and which apps are fine to connect. Explain that dots act in the background.
-
Update offboarding. Add a step to pause or delete a departing member’s dot.
Edu admins have one extra question before the dots admin preview ends. If students share the workspace, decide whether the default should include them. Custom roles can keep dots to faculty and staff.
How Anthropic and Google handle agents
We covered Anthropic’s proactive agent app in What Is Claude Duet? and Google’s work agent in What Is Gemini Agent?. For other features that turned on by default this fall, see Gemini Defaults in Google Workspace and ChatGPT for Word Turns On by Default.
Frequently Asked Questions
When does the dots admin preview end?
OpenAI’s help article says it runs from November 4 through November 18, 2026. From November 18, dots are available by default in eligible Enterprise, Edu and Healthcare workspaces unless an admin opts out.
Will every member get a dot when the dots admin preview ends?
No. OpenAI says enabling access does not create dots for members. Each member creates their own, and each can create one personal dot.
How much do dots cost on ChatGPT Enterprise?
OpenAI lists flat rates mapped to the type of work, or scaled consumption. It has not published dollar rates as of November 6, 2026. Ask your OpenAI rep.
Can admins turn off proactive research?
Not on its own. OpenAI’s admin article says there is no separate setting for it. Your controls are dots access and which apps members can connect.
Can a dot use a member’s own computer?
Only if an admin allows local computer access and the member connects their computer. OpenAI says that permission is off by default for Enterprise.
Can Enterprise members text their dot?
No. OpenAI says phone messaging through iMessage, RCS or WhatsApp is unavailable for Enterprise. Slack works with the right permissions, and Teams has limited availability.
Does OpenAI train on dots content from our workspace?
OpenAI’s launch post says it does not use content from Business, Enterprise or Edu workspaces to improve its models by default.
What happens to a dot when a staff member leaves?
Removing the member revokes their access but does not delete the dot. A workspace owner can delete it. Separate scheduled or delegated tasks need their own review.
