Meta tests facial recognition to combat scams and restore profiles

Contents

Meta Platforms plans to use facial recognition to take action against scammers who advertise with well-known personalities or pretend to be such. This is because advertising campaigns on Facebook, for example, are checked automatically using artificial intelligence methods such as machine learning. Facial recognition is now being used as a test. The Facebook group is also testing this for lost access to its own profiles. In future, video selfies will also be sufficient for identification and thus for the recovery of user accounts if the recognized face matches pictures of the profile.

The company does not have a good track record with facial recognition. For years, Facebook had biometrically analyzed the faces of more than a billion people – often without consent. After protests by civil rights activists in 2021, Meta deleted Facebook's facial recognition profiles. The company was sued anyway. In 2022, Meta compensated users in Illinois for unauthorized facial recognition on Facebook, which amounted to 650 million US dollars. It was more expensive in Texas. There, facial recognition cost Meta 1.4 billion dollars.

Fighting scams by comparing images

The Facebook group now wants to use facial recognition productively for positive purposes. One starting point is the fight against scams on its own platforms. Scammers often use images of well-known personalities to advertise their offers and generate clicks (celeb bait). Such offers are often used to steal personal data and even money. This violates the terms of use, writes Meta, and would harm users.

As advertising on the Meta platforms is not checked and approved by humans, but automatically, some scamming methods are not recognized. The company now wants to use facial recognition to better filter out abusive campaigns involving celebrities. Meta has successfully tested this with a small group of celebrities and will be expanding it in the coming weeks. The people concerned will be informed and can decide for themselves whether they want to take part (opt-out). Meta promises to delete all data collected during the facial recognition process once the check has been completed.

This also applies to the verification of scammers who claim to be a known personality. Some would create corresponding profiles on Meta's platforms and advertise products and investments, for example, in order to persuade victims to send data or money. With a facial recognition check of profile pictures, such methods would be recognized more quickly if these pictures can be compared with the real profiles of these people on Facebook or Instagram. This process is to be tested in the near future.

Facial recognition for login recovery

Facial recognition should also help to restore lost access to Facebook or Instagram, for example if the user has forgotten their password, lost their device or someone else has gained access. If a user account has been compromised, Meta has so far required the presentation of an ID or other official document with the user's name.

The company is now testing video selfies to verify the identity of the user. This is comparable to facial recognition for unlocking a smartphone. As soon as the video selfie has been uploaded, it is encrypted and stored securely. The video is never made publicly accessible and is deleted after verification – regardless of whether the identification was successful or not.

British and EU citizens excluded

Although the tests with facial recognition are carried out worldwide, they are not carried out in the UK and the EU. In these countries, stricter data protection laws require explicit consent for such methods. "We are in dialog with the UK regulator, policy makers and other experts as the tests progress," explains a Meta spokesperson. "We will continue to gather feedback from experts and make adjustments as the features evolve."

Facial recognition for security purposes and to combat scams could be part of Meta's strategy to persuade European legislators to open up data protection regulations, writes Techcrunch. After all, the Facebook group also wants to collect more data in order to underpin its ambitions in artificial intelligence, for example.

(fds)