OpenAI has launched a centralised File Library in ChatGPT, giving paid subscribers persistent document storage accessible across multiple sessions for the first time.

The feature rolled out on 24 March 2026. It automatically saves documents, spreadsheets, presentations, PDFs and images uploaded during chats. Files are accessible from a dedicated sidebar panel at chatgpt.com.

Previously, files were tied to individual conversations and lost when a chat was deleted. Under the new system, files persist in user accounts until manually removed.

The Library is currently available to ChatGPT Plus, Pro and Business subscribers only. Free users are excluded.

The rollout excludes the European Economic Area (EEA), Switzerland and the United Kingdom - reflecting ongoing regulatory pressure on AI data handling in those regions. OpenAI has not provided a timeline for availability in those markets. Australian users on eligible paid plans are included in the rollout.

The feature is web-only at chatgpt.com. Mobile apps for iOS and Android support recent file access and file search, but do not yet offer full Library functionality.

Individual users receive up to 10 gigabytes of storage. Organisation-level accounts under Business and Enterprise plans receive up to 100 GB.Per-file limits include a 512 MB hard cap, a 2 million token limit for text and document files, and a 50 MB cap for spreadsheets. Images are limited to 20 MB.

Files can be filtered by type - documents, spreadsheets, presentations, PDFs or images - and downloaded individually or in bulk.

Data governance implications

The Library creates new considerations around AI-based file storage and retention.

OpenAI states that for Business and Enterprise customers, data is not used to train models by default. The platform carries SOC 2 Type 2 compliance and supports GDPR and other data protection frameworks.

However, once a user manually deletes a Library file, OpenAI retains the data on its servers for up to 30 days. Permanent deletion occurs only after that window closes. OpenAI has not specified the reason for this retention period.

Research published in 2025 found that sensitive data makes up 34.8% of employee ChatGPT inputs - up sharply from 11% in 2023.

The Library’s persistent storage model extends the risk of such incidents. An uploaded file is now retained indefinitely in a user account - not limited to a single session.

Analysts noted that Business and Enterprise customers can apply administrative controls and retention policies. However, they advised treating the Library as a convenience layer above core document systems - not a replacement for regulated storage.

Comparison with competing platforms

The File Library narrows a gap that rival platforms have addressed for some time.

Anthropic’s Claude Projects feature offers persistent file storage across conversations for paid subscribers. Microsoft Copilot integrates natively with SharePoint and the Microsoft 365 suite, accessing organisational document repositories without requiring separate uploads. Google Gemini connects to Google Drive, enabling retrieval from existing cloud storage.

A 2026 enterprise comparison noted Microsoft Copilot’s defining advantage is deep integration with Microsoft 365. This includes sensitivity label enforcement - restricting access to content based on user permissions and security classifications.

A 2026 analysis by IntuitionLabs found 78% of Global 2000 organisations now run OpenAI models in production. Many have adopted multi-platform strategies, using two or more AI providers concurrently.

OpenAI’s official documentation on file storage and Library settings is at https://help.openai.com/en/articles/20001052-file-storage-and-library-in-chatgpt