Facial recognition has always been problematic from a privacy standpoint.
By utilising the technology, all it takes to identify somebody is taking a photo of them.
But now, researchers have introduced Fawkes, an AI tool that is certain to make this much harder.
Named after the iconic Guy Fawkes masks from the movie V for Vendetta, the algorithm alters your photo ever so slightly to trick facial recognition systems without making a difference to the human perception.
In a way, it is almost like you are wearing an invisible mask.
From a technical standpoint, Fawkes makes subtle changes to your photos so that facial recognition systems see you as a different person.
This makes it a suitable solution for protecting the photos you share online from facial recognition such as that run by Facebook.
Fawkes boasts a 100% success rate against the following state-of-the-art facial recognition services:
- Rekognition (by Amazon)
- Azure Face (by Microsoft)
- Face++ (by Megvii)
Earlier this year, a paper on the algorithm was published by Ben Y. Zhao, Shawn Shan, Emily Wenger, Haitao Zheng, Jiayun Zhang and Huiying Li.
Last month, the authors decided to release Fawkes as free software that anyone can download.
The software is available for Mac and Windows.
So far, it has amassed over 100,000 downloads.
It takes a couple of minutes to process each image, but Fawkes is generally user-friendly.
Fawkes is a major step forward for the protection of privacy, but it needs some form of mass adoption for it to reach its full potential.
Another problem is that it will not undo whatever data facial recognition algorithms have already gathered in the past.
Still, Zhao believes that cloaked photos may very well outnumber the amount of uncloaked photos if people start using it.