Not all facial recognition is successful at telling the difference between the genuine face of an authorised user and, say, a photograph of the user.
Someone who has realised that in recent days is Matt Carthy, an Irish politician serving as an MEP at the European Parliament.
With new elections for Europe just around the corner, Carthy can’t be the only politician liberally distributing photographs of his smiling face through the letterboxes of potential voters.
But in Carthy’s case it wasn’t a stranger that was able to subvert the facial recognition on his HP laptop, but instead his kids.
So, I was wondering why the battery on my laptop was running down every time I left it at home.
Turns out the kids have been using my election leaflets to get through the facial recognition lock…
I’m not sure whether to be proud by the wit or concerned by the sneakiness? pic.twitter.com/rtDsuNRB8B— Matt Carthy TD (@mattcarthy) April 23, 2019
The problem with using your face or your fingerprints as the magic key that will unlock your device is that these are things which simply aren’t secret.
If your laptop or smartphone isn’t sufficiently adept at telling the difference between you and a mugshot of you, or your fingerprints and a 3D print of your fingerprint, you might be putting your sensitive data and privacy at risk.
Not a great advert for HP’s facial recognition.
Back in the 1980s there was a lot of research into signature verification as a means of proving identity. Many devices were used, the most successful of which was a "ballistic" pen. But in the end they all failed, for a very simple reason: if you set the system tolerances to ALWAYS recognize the right person, it will also sometimes recognize the wrong person. If you set them so that the system NEVER recognizes the wrong person, it sometimes also declines to recognize the right person. Neither is acceptable, so the research ended. Doubtless facial recognition suffers from the same dilemma.