How to Share ChatGPT Library Folders With a Team

Your team can now keep an approved reference folder inside ChatGPT, give colleagues Viewer or Editor access, and let them use those files in conversations without sending another copy. The important change is ownership: a file added to a shared folder belongs to the folder owner and stays there if the uploader later loses access.

The short answer

Share a ChatGPT Library folder from its sharing dialog, invite specific people or make it available across your workspace, then assign Viewer or Editor access. Recipients find the folder under Shared with me and can use its files in ChatGPT conversations. That is the workflow OpenAI added on September 9, 2026.

Here is the clean setup:

1. Open Library on the web and open the sharing dialog for the file or folder you want the team to use.
2. Invite the specific people who need it, or choose workspace-wide access when everyone in that workspace should receive it.
3. Assign Viewer to people who only need the reference set. Reserve Editor for the smaller group responsible for maintaining it.
4. Ask one recipient to open Shared with me and use a file in a fresh conversation. This tests discovery, access, and the real working path.
5. Before removing anyone, review what they uploaded. Their files stay with the folder owner, while the uploader loses access.

OpenAI names the two roles but has not yet published a detailed action-by-action matrix for them. Treat Viewer as the least-privilege default, use Editor for maintainers, and confirm the controls shown in your own sharing dialog before putting sensitive material into the folder.

What changed, in plain language

Library sharing turns ChatGPT's file store into a shared reference shelf. The shelf owner controls the keys. Teammates can be invited to the shelf, given one of two access labels, and use what is there while working with ChatGPT.

That is different from passing attachments through Slack or email. Those copies drift. A shared Library folder gives the team one place to return to, which makes a current brief, policy, research pack, or product sheet easier to reuse.

There is a documentation wrinkle worth knowing. The dated release note introduces sharing, while the older general Library guide still says Library does not introduce collaboration. The September 9 release is the newer authority. It does not yet give a separate plan-by-plan rollout table for sharing.

The underlying Library is available to Free, Go, Plus, Pro, and Business users, plus Enterprise, Edu, and Healthcare workspaces. Library itself is documented on the web. If your account does not show a sharing control, the current official pages do not provide enough detail to promise when your plan, region, or workspace will receive it.

Viewer or Editor: use fewer Editors

The safest policy is simple: most people should view the reference set, and a named owner plus one backup should maintain it.

This is a governance recommendation, not an undocumented permissions claim. OpenAI has announced the role names but has not yet listed every action each role allows.

The ownership rule matters more than the share button

A file uploaded to a shared folder belongs to the folder owner. ChatGPT still records who uploaded it, but the folder owner keeps the file when that person's access is removed. The uploader then loses access to the file.

Think of a teammate placing a binder in a company records room. Their name remains on the intake card, but the binder becomes part of the room owner's collection. Taking back the teammate's badge does not remove the binder.

That has three practical consequences:

• Put durable team folders under a durable owner, not a short-term contractor or a person likely to leave.
• Tell contributors about the ownership rule before they upload client, licensed, or personal material.
• Add a folder review to offboarding. Removing access is not the same as removing what the person contributed.

The budget line this can change

The business case is coordination time, not a cheaper ChatGPT seat. Use this quick test:

people × file hunts or resend requests per week × minutes each = coordination time

For an eight-person team making two five-minute searches or resend requests each week, that is 80 minutes of avoidable handling. A shared reference folder will not erase all of it, but it gives you a measurable before-and-after workflow: count the requests for two weeks, create the folder, then count again.

There may also be an add-on budget to inspect. AI Toolbox, an adjacent browser extension that organizes ChatGPT conversations, lists a team plan from $12 per seat per month. Eight seats would be $96 a month. Native Library sharing does not replace its chat folders, search, or export features. But if you bought a separate tool mainly to circulate reference files, test the native folder before renewing that part of the stack.

This capability does not remove ChatGPT plan or workspace seat costs. OpenAI's release note also does not describe public or anonymous Library links. Treat it as controlled collaboration inside ChatGPT, not a free external portal.

Seven team workflows worth trying

The teams that gain most are the ones that repeatedly reuse approved source material.

The first workflow is the strongest. Marketing teams combine high file reuse with costly approval mistakes, so a controlled Viewer-heavy folder can remove more friction than a general pile of shared documents.

Library, Project, Site, Drive, or a local folder?

Choose the surface based on what the team needs to share.

Do not use a shared Project merely because the word "folder" feels similar. Projects bring chats, instructions, and memory into the collaboration boundary. Library sharing is the narrower choice when the job is reusable files.

What is worth building around this

1. Best bet: a Library governance onboarding package

Sell a fixed-scope setup to marketing, sales, support, and agency teams: folder taxonomy, owner selection, a Viewer-versus-Editor assignment sheet, an upload policy, and an offboarding audit.

The demand is narrow but commercial. The keyword dataset reports 70 monthly searches for chatgpt team workspace, with an average cost per click of $12.52. The live questions also ask whether other Business members can see a person's chats, which points directly at access-boundary anxiety.

The smallest sellable version is a discovery call, one pilot folder, a permission matrix, a short team walkthrough, and a review after two weeks. No integration is required. The catch is that OpenAI has not documented a Library-sharing API or a complete role-action matrix, so the offer must remain a human-reviewed operating system rather than promise automated enforcement.

This is the strongest opportunity because ownership and offboarding are not solved by the share dialog. They are policy problems, and teams pay to make policy concrete.

2. A reference-file migration audit for folder-extension users

Help teams decide which material belongs in native Library folders and which conversation organization still needs a browser extension. The dataset reports 320 monthly searches for chatgpt folders extension, and AI Toolbox prices its ChatGPT module at $9.99 per month or its team product from $12 per seat per month.

The MVP is an inventory worksheet, a keep-or-move decision tree, a target folder map, and a supervised migration of the highest-value reference files. The catch is important: those extensions organize chats, while Library sharing handles files and folders. There is no documented bulk migration path, so a promise to replace every extension feature would be false.

Limits and the honest take

Native sharing is useful, but the documentation is still thinner than the workflow deserves.

• There is no published sharing-specific plan matrix yet. General Library access spans Free through enterprise workspace plans, but that does not prove identical sharing availability everywhere.
• Viewer and Editor are named, but their full action matrix is not published. Verify the live dialog before relying on a permission assumption.
• Named people and workspace-wide access are documented. Public or anonymous Library links are not.
• Library is documented on the web. Temporary Chat uploads do not enter Library.
• Workspace owners can disable automatic referencing of Library files. Members can still browse, search, open, or attach those files manually.
• The general file rules still apply: 512 MB per uploaded file, 2 million tokens for text and document files, about 50 MB for spreadsheets depending on row size, and 20 MB per image.
• Published storage is 500 MB on Free, 4 GB on Go, 20 GB on Plus and Business, and 100 GB on Pro. Enterprise, Edu, and Healthcare files follow workspace retention policy.
• No Library-sharing API or bulk governance controls are documented in the pages available at launch.

The honest take: use this now for a controlled source-of-truth folder, but do not redesign your whole knowledge system around a one-day-old sharing layer. Start with one folder whose owner, editors, viewers, and offboarding rule are obvious.

Can I share ChatGPT projects with others?

Yes. Shared Projects are a separate ChatGPT feature. They combine chats, files, custom instructions, and project memory, and their roles are Chat and Edit. A shared Library folder is narrower and uses Viewer and Editor labels.

Can you attach folders to ChatGPT?

For a connected Google Drive folder, OpenAI's Library guide says you can select the folder and ask ChatGPT to work across its files. For a newly shared Library folder, the release says recipients can find the shared item and use its files in conversations, but it does not say the whole folder attaches as one unit. A folder on your local computer is not automatically attached.

Can I share my ChatGPT chat with someone?

Yes, but chat sharing is not Library sharing. A shared chat link exposes that conversation according to the chat-sharing controls. A shared Library item gives access to the selected file or folder, while a shared Project can include project chats, files, instructions, and memory.

Can other members see my chats in ChatGPT Business?

Library sharing is described as access to the files or folders you share, not blanket access to private chat history. A shared Project is different because its members can see and interact with that project's shared chats and files. Check the boundary before choosing the surface.

If you want a permission and handoff workflow designed around your team's tools, see AI automation services.