Legal Area: Virtual assistant: smart or intrusive things?

Key-words: Virtual assistant – privacy

Virtual assistants are gaining success, reaching very important goals for the new HiTech border. They work through vocal commands, performing not only their basic speakers’ skills but also interacting with users in their daily routine: domotics, weather forecast, news update driving fucnions (see Huracán Evo, Lamborghini).

How can Amazon, Alexa or Google Assistant satisfy all our requests? Do they always listen to us?

This kind of device should start up with an activation word, the vocal recognition system decodes the user and find answers in its own back-end. However in some cases, unfortunately, they are already active by default.  Behind Virtual Assistants, there are real people hearing users requests.  Through “data annotation” operations, they listen to conversation clips and note down what is said to fill language comprehension gaps and make their virtual assistants even smarter.

Data processing and profiling processes are only two of the numerous issues related to user's privacy.

The fact raised lots of discussions for Amazon: we talk about thousands of people with the scope of listening to others’ conversations.

In this regard, Amazon explained that users can disable the use of the recordings default option. They analyse extremely reduced samples of these recordings, ensuring maximum confidentiality and only users consent can authorise the data processing

These answers Amazon provided, were not enough for users.

Consequently, the Company made some changes on new devices: new series 8 is equipped with a microphone-camera deactivation button and a clear visual indicator showing when audiovisual communications are forwarded to Cloud.

For previous devices’ users, it is possible, instead, to safeguard users privacy with few simple steps to make sure they’re not overheard:


Source: Sole24ore e LaRepubblica.



To read the PDF click here.