ChatGPT Library Comes to Enterprise, Edu, and Healthcare: A Persistent File Hub with Admin Controls
ChatGPT Library gives Enterprise, Edu, and Healthcare workspace members a dedicated, policy-compliant space to store and reuse uploaded files.
Anyone who has used ChatGPT in a professional context knows the frustration: you uploaded that research PDF three weeks ago, had a great conversation about it, and now you need it again. Good luck finding it. You scroll through a list of chats with names like “Untitled” and “Data analysis help” hoping something jogs your memory. More often than not, you just re-upload the file and start over.
OpenAI has now addressed this directly. ChatGPT Library is rolling out to Enterprise, Edu, and Healthcare workspaces, giving members a dedicated, persistent space to store and retrieve the files they upload or create in ChatGPT.
What Library Actually Does
Library automatically saves files the moment you upload them to a conversation. Documents, spreadsheets, presentations, PDFs, images — they all go into a personal Library, accessible from the left sidebar on the web. Files created during a conversation (say, a CSV generated by ChatGPT from your data) are saved there too.
From that point on, you can browse, search, filter by file type, and attach any stored file to a new conversation without uploading it again. The file is just there, ready to use.
Critically, Library storage is separate from daily attachment and chat limits. Deleting a conversation does not delete the files it contained. Files persist in your Library until you choose to remove them manually, at which point OpenAI removes them from its servers within 30 days.
Storage Limits Worth Knowing
Individual users get up to 10 GB of Library storage. Organisation-level accounts on Business and Enterprise plans get up to 100 GB. Individual files are capped at 512 MB, with a 2 million token limit for text and document files, a 50 MB cap for spreadsheets, and a 20 MB limit for images. Bulk downloads are supported, so you are not locked in if you want to pull files out.
The Admin Controls Matter Here
For Enterprise and Edu admins, the more interesting part of this release is probably not the storage itself — it is the controls around how ChatGPT uses those stored files.
Workspace owners can decide whether ChatGPT automatically references Library files when members send a prompt. With automatic referencing on, ChatGPT may proactively draw on stored files to inform its responses, even when a file has not been explicitly attached to a conversation. With it off, members can still browse, search, and manually attach Library files — they just need to choose the file themselves rather than having ChatGPT reach for it automatically.
This is a meaningful distinction for organisations that want predictability in how the model behaves. If staff are asking ChatGPT general questions, you may not want it silently pulling context from a financial model someone uploaded two months ago.
For Healthcare workspaces specifically, automatic referencing is off by default. That is a sensible baseline given the sensitivity of the data environments those teams typically operate in.
Retention Policies and Compliance
Library files follow your workspace retention policies, which is the piece that makes this genuinely usable for regulated industries rather than just convenient.
OpenAI has also added Library-specific Compliance API endpoints. Compliance admins can export and delete Library files programmatically through the same compliance platform used for other Enterprise and Edu data management. For teams that already have data governance workflows built around ChatGPT Enterprise, this extends that coverage to Library without requiring a separate process.
The platform maintains SOC 2 Type 2 compliance, and as with the broader Enterprise offering, data is not used to train OpenAI models by default.
What This Does Not Do
Worth being clear on a few things Library does not include. There is no file sharing between workspace members, and no collaborative or multiplayer access to files. Library is personal storage, not a shared drive. If you are looking for something closer to SharePoint or Google Drive with shared folders and permissions, this is not that.
Library is also currently web-only.
How It Fits the Broader Picture
ChatGPT has traditionally been a stateless tool — each conversation starts fresh, and anything you uploaded existed only in the context of that chat. Library is a meaningful step away from that model, moving toward something closer to a personal workspace where your files accumulate and remain accessible over time.
Anthropic’s Claude Projects and Microsoft’s Copilot integration with Microsoft 365 have offered versions of persistent file access for a while now. Library brings ChatGPT Enterprise closer to that standard, with the added layer of workspace-level policy compliance and admin oversight that those environments require.
What This Means for Your Team
For members using ChatGPT regularly, the practical benefit is straightforward: less time hunting for files, no more redundant uploads, and a cleaner way to carry reference material across multiple conversations. If your team regularly feeds ChatGPT the same set of documents — a brand guidelines PDF, a product spec, a data dictionary — those files now live in one place and can be attached to any new conversation in a couple of clicks.
For admins and compliance teams, the combination of retention policy compliance, Compliance API support, and the automatic referencing toggle gives you the oversight you need to deploy this confidently. The Healthcare default of automatic referencing off is a good signal that OpenAI has thought about the sensitivity gradient here rather than shipping a one-size-fits-all feature.
Library is rolling out now across Enterprise, Edu, and Healthcare workspaces, including those in the European Economic Area, Switzerland, and the United Kingdom.