Beware of AI assistants

Artificial intelligence (AI) has been used by a wide audience for a few years now, usually in the form of a chatbot providing answers to all kinds of questions. The security and privacy risks of these tools are not insignificant, as the questions asked can reveal a lot about the private lives of those who ask them.

A new trend has emerged more recently, ‘AI agents’ that perform actions on behalf of users: reading and writing emails, ordering items online, booking trips, organising hard drives, etc. The Muse software developed by Meta (owner of Facebook, Instagram and WhatsApp) is an example of this, and has been downloaded millions of times in the United States.

Such software, which may seem attractive at first glance because of the services it provides, requires access to a great deal of data stored on a computer or smartphone in order to function: email inboxes, calendars, messages, address books, hard drives, information displayed on the screen, etc. It is therefore essential to only grant the necessary access after carefully studying the developer’s commitments regarding how your data will be used. If in doubt, it is better to refrain from using the software in order to avoid having your digital life sucked up by a company that could be not very scrupulous.