Logo and page links

Main menu


Archiving is a field associated with great efficiency potential by digitization. Artificial intelligence is expected to play a significant role in this. That raises quite a few privacy issues.

The public sector claims they with the current manual processes are unable to comply with the requirements of, for example, the Public Administration Act and the Archives Act. In addition, there are large gaps in record keeping and archiving, because a lot of archival information "remains" in the employees' inboxes. The project aims to clarify how an AI solution can be used for automatic archiving of e-mails.

Privacy and archiving are often perceived as opposites. The stereotype of the archivist will save on everything, while the privacy stereotype requests everything to be deleted. It is admittedly more nuanced than that. Saving the right documents is a good privacy practice. And the archivist will never have unlimited storage capacity, nor does he want the truly archival worthy to drown in the heaps of less relevant documents.

But choosing what to archive and not isn’t easy. And it is extremely resource intensive. So Simplifai wants to explore the possibility of using machine learning to identify and suggest which e-mails and associated documents are considered archival material.

The sandbox process

Simplifai has already participated in the National Archives' sandbox, where they developed DAM, a solution for automatic archiving of e-mails using artificial intelligence. The National Archives' sandbox concluded that DAM is legal used on public mailboxes, such as post @ addresses (ie not for the employees' individual email addresses).

But for DAM to work as intended, as a tool for employees' e-mails as individual DAM, the underlying AI model must be trained with all types of relevant data that may appear in the correspondence, including personal information and special categories of personal data. Whether this can be done within good privacy, and if so - how, is the main issue Simplifai brings into the sandbox.

The project plan is just around the corner, and will be published when it is ready.