ChatGPT Turns Your Files into Digital Gold, But at What Cost?

OpenAI has recently introduced a new feature for its paying subscribers, called Library, which transforms ChatGPT into a true personal cloud. This new addition allows users to automatically store all files sent during conversations with the AI, whether they are PDFs, spreadsheets, presentations, or images. The stored files can be reused from one conversation to another, providing undeniable convenience, but raising important questions about personal data management.

Default Storage Without Prior Consent

The Library feature is available only on the web for subscribers of the Plus, Pro, and Business plans. Free users, as well as those residing in the European Economic Area, Switzerland, and the United Kingdom, cannot benefit from it. OpenAI has not provided a timeline for extending this feature to these regions, where the GDPR imposes strict restrictions on personal data management.

The default operation mode of Library is noteworthy: file saving is automatically enabled, without user intervention. Files remain stored even after a conversation is deleted, requiring manual deletion in Library. Additionally, OpenAI retains data on its servers for 30 days after deletion. Storage limits include a maximum of 512 MB per file, 10 GB per account, and 20 MB per image.

Towards an Integrated Workspace

With Library, ChatGPT continues to evolve into an integrated workspace. Following the introduction of Projects and connectors for Google Drive and Slack, this new feature strengthens users' engagement within the OpenAI ecosystem.

Exploiting Private Data: A Strategic Challenge

Beyond user convenience, OpenAI's strategy aims to exploit private data as a new resource for training its AI models. As the AI industry faces a shortage of usable public data, Library provides access to high-quality private data. For these files to be used in improving the models, the option “Improve the model for everyone” must be activated by the user.

This approach presents a dual benefit for OpenAI: it increases users' dependence on its service, making any migration to a competitor costly, and it enriches its AI models with data provided for free by users. The exclusion of Europe reveals that this model does not comply with the strict standards of the GDPR, highlighting the tensions between technological innovation and personal data regulation.