How many times have you uploaded photos to Facebook and thought, “how does this app know that pic is of weird Uncle Charlie?” Perhaps you wondered briefly and then went about the business of uploading your child’s birthday party pics, or last weekend’s barbecue.
However, the reality is that Facebook is using a very powerful tool known as facial recognition software. In fact, many say that Facebook’s actual facial recognition program is much more powerful than that used by the FBI to catch every ne’er do well from here to Helsinki.
But what if someone were able to secure every picture associated with Facebook? Other than a certain violation of your privacy – what harm could there be? The answer might surprise you. Why? Because if everyone had access to all of the social media photos on apps such as Facebook, Instagram, Twitter, Musical.ly, SnapChat, etc, it would mean they’d have a window into your identity – without having ever met you.
For example: Government uses facial recognition technology to catch the bad guys. However, 26 states use this same technology for everyone with a driver’s license. This allows government agencies to easily identify you – and identify anyone that has stolen your identity.
Maybe we’re ok with our federal and local governments being able to identify us under the auspices of preventing identity theft. But what about social media companies? Should they have access to such power?
As suggested, Facebook’s technology is better at identifying people through photos than the FBI. This is in part because Facebook has 1.65 billion users – each of whom post hundreds – if not thousands of pics each year. Therefore, their data sets are better and facial recognition learning becomes enhanced. This is called Machine Learning.
But can someone steal your photos on social media? Today, if a user’s account does not have the proper privacy settings, those photos can be compromised and used for just about any purpose. In fact, you cannot secure your profile picture on any social media platform, They are open to the public.
To our knowledge, no one has been able to hack Facebook’s entire collection of photos and associated metadata. However, someone has been able to access a dataset containing 40,000 images of both men and women from Tinder. Given that most of these photos are not of “the big 3″ i.e., birthdays, bar mitzvahs or barbecues – the stakes are little higher for those whose images grace the digital vaults of Tinder.
Stuart Colianni wrote a program to compile Tinder photos, intending to use them for machine learning research. His rationale: “Why not leverage Tinder to build a better, larger facial dataset?”
For his research, he added folders containing the photos to Google’s Kaggle, a service that provides programmers the ability to experiment with artificial intelligence algorithms. Such algorithms can be focused on large photographic datasets to perform facial recognition tasks.
Well as one might imagine, this upset the folks at Tinder – but not half as much as the college students, married doctors, Lawyers, CPA’s and other professionals and non-professionals whose alter-egos “were looking for love in all the digital places.”
So what’s the big deal? In short as was reported in a recent Vocativ article, “… strangers can use them to catfish others — the act of posing as a person to lure another. Internet strangers can also reverse search nameless photos and potentially find who they belong to as well.”
Again, it’s hard to compare Tinder’s User Stats to that of Facebook’s 1.65 billion users. But given the perspective mentioned by Vocativ above, imagine the damage that could be inflicted on Facebook users if every such photos became part of a sophisticated, AI, facial recognition database.
We must come to grips in this generation of algorithms — that our faces are nothing more than mathematics. In fact, the math that makes up our face never changes. As such, if you have sophisticated facial recognition technology — your high school picture could be scanned and subsequently find you on any number of social media platforms.
Surprised by all of this? Most people are. However, flying under the radar for most of us is Google’s Vision and Amazon’s new Image Recognition service called Rekognition.
Amazon’s service was launched in 2016 and according to their website, “Rekognition’s API lets you easily build powerful visual search and discovery into your applications. With Amazon Rekognition, you only pay for the images you analyze and the face metadata you store. There are no minimum fees and there are no upfront commitments.”
Although this service won’t help you hack Facebook or other social media apps, it will help you identify people in photos that you might take or have found elsewhere. There is a free tier that will at help you understand how powerful facial recognition is — and where this technology might take us in the future.
Not everyone wants to pay for such services. In that case there are reverse image look up sites such as Tin Eye that can find identical images on the web by scraping social media sites to unearth the exact images that you use for your search.
This is just the beginning.
Fortunately, the scrapping of 1.65 billion user photos on Facebook did not happen this week. However, there are 40,000 Tinder users wondering what embarrassing images of them might be floating in the ethers of the Internet.
But you’re still thinking… “how did Facebook know that was Uncle Charlie?”