ChatGPT Business no longer forgets your files

OpenAI expanded ChatGPT's File Library to Free and Go users this month, and extended access to the EEA and UK for the first time. Files now persist across sessions, with storage limits set by plan tier.

The bigger story is structural. Browser-based AI has had a quiet weakness for years, and ChatGPT just closed it.

What changed in May

ChatGPT's File Library first launched in March 2026. At that point it was available to Plus, Pro, and Business users only, and not in Europe. The May update extends it on both fronts. Free and Go users now have access, and the EEA, Switzerland, and UK are included.

Storage tiers are set by plan:

• Free: 500MB
• Go: 4GB
• Plus and Business: 20GB
• Pro: 100GB

Files you upload, and files ChatGPT generates, are saved automatically to a Library tab in the sidebar. You can browse, search, filter, and attach them to new chats via the composer's "add from library" option.

You can also ask ChatGPT about past uploads directly. Questions like "what was in the document I shared yesterday" work in any chat.

Files uploaded in Temporary Chat are not saved. The full Library view is web-only for now. Mobile gets the recent files and search functions but not the dedicated tab.

Why browser AI had this gap, and desktop AI didn't

Browser-based AI tools have a structural problem desktop AI workspaces don't share. The browser tool lives in a tab, separate from where your files actually sit. To use it, you bring files to the AI: download, upload, attach, repeat.

Desktop AI workspaces work the opposite way. Tools like Codex, Cowork, or Claude Code sit inside your file system. They see your files where you keep them, without you copying anything across. The AI comes to your files, not the other way around.

Both approaches have trade-offs. Browser tools are simpler, work on any device, and need no install. Desktop tools handle real working files natively, with no copying step.

For most teams across the UK SME market, the browser is where AI happens. That's where staff already work, and that's where the bulk of daily AI usage sits.

Until now, browser AI carried the drag of repeated uploads and lost files. ChatGPT's File Library closes that gap. The browser AI workspace now keeps your files between sessions, the same way a real workspace does.

What this means for ChatGPT Business teams

For business teams on ChatGPT, the practical change is bigger than the announcement makes it sound. The 20GB Business storage tier is large enough for sustained team use. Ongoing projects, reference documents, templates, and shared assets stay accessible across sessions and across team members.

The follow-on effect lands on adoption. AI adoption is 80% people and culture, 20% technology. Friction kills daily use; small UX wins drive it.

When staff don't have to re-upload the same client brief every morning, AI becomes part of the workflow. It stops being a tool you visit.

The shift from "AI is something I open" to "AI is somewhere I work" happens through dozens of small changes. This is one. Each removes a reason to switch away from the AI tool. Each adds a reason to come back tomorrow.

If you want a clear read on where your team sits with AI adoption, and which workflows belong in which tool, take the AI readiness assessment.

The wider workspace play

File Library doesn't land in isolation. OpenAI has been building toward a workspace platform for months. Projects launched in February 2026 to give chats a shared context. Excel and Google Sheets now have ChatGPT sidebars. The Atlas browser sits AI inside browsing. A consolidated superapp has been signposted in OpenAI's roadmap.

Persistent files are the foundation that makes the rest viable. A workspace needs continuity. Documents that vanish at session end don't support real work. Library makes ChatGPT a place where work accumulates, not a place where work is performed once and lost.

This matters for SME buyers comparing platforms. The choice is no longer "which AI model do I use", it's "which workspace do I commit to". ChatGPT, Claude, Microsoft Copilot, and Google Gemini are pulling in the same direction. Each is moving from chatbot to workspace. The faster a buyer reads that shift, the better the platform decision they make.

What SME leaders should think about now

Three practical implications follow for any SME using ChatGPT Business.

First, file governance changes. Files now persist by default. Anything staff upload sits in the Library until someone deletes it. That includes client documents, financial data, draft contracts, and anything else flowing through chats.

A clear policy on what should and shouldn't be uploaded becomes more important, not less. Temporary Chat exists for files that shouldn't persist; staff need to know how and when to use it.

Second, ownership and access need a conversation. The Library is tied to user accounts. When someone leaves, their Library leaves with them, unless your governance and offboarding cover it.

For team workflows that depend on shared files, Projects is the better home. For individual day-to-day work, Library is fine. The line between the two should be deliberate.

Third, this is an opening to revisit where AI work happens. Some workflows belong in browser AI. Others belong in desktop AI that sits inside the files. The right answer depends on the task, the data sensitivity, and where staff already work.

The wrong answer is to default to one tool because it was the first one installed.

What this changes for AI adoption

AI adoption is 80% people and culture, 20% technology. The technology side of this update is small: a Library tab, a few storage tiers, a search box. The people side is larger. It removes one of the daily reasons staff drift away from a tool.

The gecco view: small UX changes like this drive bigger behaviour shifts than feature launches do. The SCALE framework's Apply phase depends on it. Friction is the silent killer of AI adoption, and every layer of friction removed is a layer of adoption gained.