Privacy and Photo:Tech

You are strolling down the beach with your 5-year-old son, happily taking pictures on your brand new  iPhone 6, sharing the best ones on your Instagram feed. The sun is glowing, you are on a well deserved vacation, your family is happy …But miles away, in a dark basement, a computer server is rapidly chirping at all incoming photos on Instagram and has selected your photos because your son’s swimsuit happens to be from a brand that has hired the services of a company that mines pictures for a living.

Next thing you know, or rather, you don’t know, your casual pictures is being analyzed for content. The location ( thank you GPS), the day, the time, the surrounding landscape, your phone make and model and anything and everything else in the photo that can gave them a clue is being analyzed .  Using your Instagram profile, it also can learn much more about you, what you like, where you go and more importantly, what you like to spend your money on. Thanks to your photo, you have just sold your son and yourself to a brand and, in the process, exposed yourself to receiving targeted advertising next time you are on-line.

While this scenario is not quite there yet, it is a possibility. While humans were the only ones capable of understanding a photograph, things are changing rapidly. Thanks to Artificial Intelligence, computers are getting much better at understanding the content of images and companies, from Google to your local start-up, on cashing in.  Could this become the new battleground for privacy advocates?

As with browser cookies, photo tech companies are not interested in collecting personal information because it has little to no value. Knowing what one individual likes is only good for that one individual. Rather, these companies are looking for trends, large clusters and similar patterns in large numbers. They want to know who are the users of their brands  and those who are not, what are their habits so they can better adapt their products and marketing.  It only works on large sets and not on an individual level because they cannot adapt their products to each and every individual’s variance of taste. They might in the future, but not for now.

Intelligence agencies, however, do want to know more about individuals and there is little doubt that FBI, NSA, CIA use photo technology to track particular individuals, their whereabouts, their likes and dislikes, their taste in clothes, food ( to poison them)  and vacation spots. But that is the fruit of social media in general and something they already had from tapping into phone conversation or intercepting emails.

In fact, there is not much that individuals reveal about themselves that should be of concern. Sure, a brand can  learn a bit more about you via your snapshots but they still cannot use any of your images without your authorisation.You are more of a walking billboard when you wear those shirts with a brand logo, or those handbags with a designers name in gold letters. And you do it for free, consciously.

However, the photo:tech industry should certainly be aware of the potential backlash that a too eager start-up might create by mining too much personal information out of images. It could trigger a nationwide resistance from privacy advocates and lead to changes in federal ( if not international) legislation that could cripple the industry. Already in Europe -in Germany mostly- Google is feeling the full impact of a privacy conscious government seeking to protect its citizen. Those rules, currently applied to Google Earth and Google search could certainly spread to image analysis in general, blocking simple tasks and at length, the growth of the whole industry.

Today, The Wall Street Journal had a thorough article on this topic, pointing the finger at two companies mining images. Tomorrow, it could be cable news, the New York Times, social media.. Could it be time for the industry to create and respect a code of conduct before it is forced to abide by too restrictive laws? It certainly wouldn’t be a bad idea.

Paul Melcher will be moderating a panel on Computer Vision at the annual PACA conference on Tuesday, October 21. You should register here

Photo by nattu