If you have Ray-Ban meta-lunchers, you must reveal your privacy settings

Meta has updated The confidentiality policy of his AI, Ray-Ban Meta glasses, giving the technology giant more power on the data he can store and use to form his AI models.

The company sent an email to Ray-Ban Meta owners on Tuesday with an opinion that the features of the AI ​​will now be activated on the default glasses, according to The penis. This means that Meta’s AI will analyze the photos and videos taken with the glasses while certain AI features are activated. Meta will also store customer voice records to improve its products, without option to withdraw.

To be clear, the Ray-Ban meta-lunes do not record and do not constantly store all around the carrier. The device only stores the speech that the user says after the “Hey Meta” alarm clock.

Meta- Privacy notice on vocal services For portable devices indicates that transcriptions and voice records can be stored for “up to a year to help improve Meta products”. If a customer does not want Meta to lead to his AI on his voice, he will have to manually delete each recording of the Ray-Ban Meta Companion application.

The change in terms is in the direction of the Amazon Recent political change affecting echo users. Last month, Amazon will perform all Echo commands via the Cloud, deleting the most user -friendly option to process vocal data locally.

Companies like Meta and Amazon are impatient to hoard these lots of vocal records, because they are training data useful for their generative AI products. With a larger range of audio records, Meta AI can possibly do a better job to treat different accents, dialects and speech models.

But the improvement of its AI is done at the expense of user confidentiality. A user may not understand that if he uses his Ray-Ban meta-lunes out of the box to photograph a loved one, the face of this person can find his way in Meta’s training data, for example. The models of AI behind these products require obscene quantities of content, and it benefits companies to train their AI on the data that their users already produce.

The hoarding of user data by META is not new. Already, Meta leads to her Llama Ai models on publications that American users share on Facebook and Instagram.