Verifying a user’s age is one of the biggest challenges the social media platforms face today. To counter this, Meta-owned Instagram is now testing a number of new age-verification methods, including having the user upload a video selfie and then having its own AI software assess their age.
Currently, when some users update their date of birth on the platform to indicate that they are over 18, they are required to upload ID. Since Instagram has encountered several cases of teens using fake IDs, they had to find a new way to verify their age. Instagram also had to confront concerned parents and well-wishers about people having to upload IDs in the face of Metas colored gradient with collecting a user’s data.
While the ID-based system will remain in place for now, Instagram is developing a number of other ways to verify a user’s age. In a test run, they ask users to find three mutual friends who can verify their ages. They call this “social vouchering” although there is clearly a lot of potential for abuse here.
The most interesting way they are going about this age verification business is by using AI. Instagram is asking some users to take a selfie video and submit it for review. An AI bot will then examine the footage and check it against a range of parameters and then ‘judge’ whether you are in fact over 18.
Instagram uses the services of an online verification company called Yoti, a company that uses AI to verify the user’s age.
Instagram says Yoti trains its AI with “anonymous images of different people from around the world who have transparently allowed Yoti to use their data and who can ask Yoti to delete their data at any time.” And for those under the age of 13, Yoti collected data with the express consent of their parents or guardians.
Instagram says that once you upload a video selfie and Yoti uses it to verify your age, the picture will not be used for anything else and will be deleted after your age is verified. However, given Instagram History of abuse of user data by parent company Metashould be careful with this approach.
This article was previously published on Source link
