Researchers leverage “Deepfakes” to give anonymity on photo sharing networks

A new role for deepfake tech

Deepfake technology doesn't have the best of reputations. Its scarily impressive ability to swap faces and voices has made it a favorite tool in blackmail, harassment, and other fraudulent activities. We've even written about deepfake phishing attacks for our blog, detailing how the tech is being used to put people at risk of identity theft.

But now, there may finally be an ethical use for this questionable technology, all thanks to a research team at Binghamton University, New York.

The team was motivated by what they saw as a need for an effective defense mechanism against the “maleficent motivations” of facial recognition and identification technology.

AI algorithms are developing at such an alarming rate that the usual counter-manipulation techniques of blurring, masking, or noise addition will soon no longer work in preserving facial identity.

As a solution, they have proposed a system called "My Face My Choice" (MFMC) that would use deepfake technology to mask a person's identity for anyone without viewing permission.

How would MFMC work?

Designed to operate within social photo-sharing networks, MFMC would give individuals control over which photos they appear in. Instead of image rights defined per image, MFMC defines them per face.

For example, when a group photo is uploaded, each individual featured within it would be able to give or deny “access” to their own face.

If they grant access, then their real likeness is there for all to see. But if they deny access, then MFMC replaces their face with that of a deepfake.

MFMC would also create different versions of the photo based on the different user-given access permissions.

The team has even taken care to ensure as little impact as possible on the original photo.

The deepfakes implement facial changes slight enough to give a negative result for any facial recognition software, but similar enough to not disrupt the content or quality of the overall image.

While the technology is still at a very early development stage, it shows great promise in giving control back to users— control over their own biometrics and their digital privacy.

As the use of facial recognition software is on the rise, we are certainly in need of any innovation that has the potential to protect personal privacy and people's identities.

Let's hope this research paper leads to development and a software roll-out. But, in the meantime, here are a few tips to help you safeguard your privacy: 6 Things You Should Never Share on Social Media