A new verification system to try to understand theage of users. Which passes (also) through afacial analysis carried out on a videoselfie. This is the strategy of Instagramannounced by boss Adam Mosseri in a post on his profile, to try to intercept minors under 13 (in Italy 14) years that according to the various laws on the subject – federal for the US and European for the EU – and the conditions of use should not open profiles and use the app controlled by Meta. But also to properly organize the user experience separating it between minors under 18 and adults.
Social networks no longer
by Riccardo Luna
When the verification request is triggered
Thus the artificial intelligence provided in this case and in a somewhat surprising way by the partnership with another company, Yoti, runs to the rescue of the platform, focused on making sure that “teenagers and adults have the right experience for their age,” as the group explained in a note. Curious that Meta is unable to carry out such a process independently. The process, more than in the registration phases, it will be triggered when a user tries to change his age to 18 or moreobviously starting from a younger age.
Why Instagram wants to become TikTok
by Francesco Marino
The other methods of age verification
Not just facial recognition: Instagram has also devised another system to certify the age of those who populate it, definitely more old-fashioned. The user who is considered to be under 13 – but who is not, of course – but also under 18 will be able to ask three of his followers over 18 to confirm his real age within three days. Finally there is the possibility of upload an identity document.
The strange bug on Instagram and how to fix it
Doubts about AI analyzing videoselfies
The most interesting aspect, and which, as always when the analysis of biometric parameters is involved, raises some concern, is precisely the videoselfie processed by AI. Yoti’s technologies are in fact able to “estimate age based on facial characteristics and share this estimate with us” adds the post on the official blog. Ensuring that “these technologies cannot link the user’s identity, only the age”. However Engadget raises some concerns about the partner chosen by Instagram for this delicate work. First, because it is already used by the German and British governments for various purposes – Instagram talks about that use in a positive way, ensuring that the company complies with all standards for privacy – and secondly, because not much is known about the actual functioning: “Not even the company knows how many and which features it uses for these analyzes” explains the US newspaper. Which also recalls thehigh error rate linked to elements such as gender, age group and skin color. The company reports a 2 and a half year margin of error.
In any case, anyone can get an idea of the capabilities of this neural network by trying it here and maybe catch these “biases” of which it would be a victim. Artificial intelligence is trained on the basis of hundreds of thousands of face images collected with the consent – so Yoti assures – of immortalized people, who can request their deletion at any time. The database of the little ones on which the AI was trained was instead compiled with the approval of the parents.
The system will serve Instagram – for now only for users located in the United States, it will be understood if the tight European meshes will allow it to be used also in EU countries – not only to prevent 13-year-olds from using the app but also to keep them away from services reserved for adults such as Facebook Datingthe section reserved for appointments, or other content that is not suitable for them as well as for prevent adults from messaging with teenagers to minors.