How AI Threatens Privacy

Human + AI > AI

By Troy Lowry

As a technologist, I’ve long been both amazed and aghast at how much information companies have on their employees. Currently, this information is vast but so difficult to synthesize and understand that it is only looked at in extreme circumstances. The speed and power of AI could change that.

Most computer tools store much more data than we realize. For instance, as I type this in Microsoft Word, it is literally storing every key stroke. This is done so I can go back to any previous version, but that data can be used for other things, as I suggested in one of my blog posts.

That’s just the very tip of a VERY big iceberg. It also stores a lot of information every time you log in, open an app, or send an email. Given the need (with a large number of skilled people for many hours), we could literally track everything you did in company systems over the course of the day and where you did it from. This data is critical to maintaining the cybersecurity of the company, as all of this information helps us verify that you are who you say you are and that you are acting in ways that make sense given your role.

In extreme cases, companies can use the information they have to see which carriers someone is connecting to us through and could subpoena them for yet more information which would likely include an exact location and the specific device you were using to connect.

Some University CIOs and I have a running joke, which is that we COULD know everything about you, every web site you visited, how much time you really spent working, how often you wear your fitness watch, even your heart rate when you walk your dog, but we don’t because you just aren’t that interesting.1

Like much humor, this is rooted in truth. If you become interesting to us, then it’s worth the time not only to get all of this data, but to follow the strict legal and access controls needed to really review them. Let’s say that somehow malware got introduced to your computer and attempted to infiltrate the entire network. Then, it’s interesting to us to go through, with a specific purpose, and figure out what happened.

AI Might Enable Much More Intrusion into Privacy

AI excels at looking at vast amounts of data and finding similarities. By corollary, it also excels at finding outliers — things or activities that don’t look or act like the usual things or activities.

New AI tools have really improved cybersecurity by exploiting AI’s ability to detect outliers. For instance, if an AI tool sees a user has accessed a server they have never accessed before and starts taking odd actions such as transferring a bunch of data or scanning all of the files, it immediately notes this and alerts cybersecurity for further investigation. This ability to sift through enormous amounts of data, correlate it, and detect outliers is a huge plus for security.

This same ability can and will be used to compromise people’s privacy. AI's capability to analyze enormous data sets in real time can easily be turned into a tool for continuous surveillance, especially in a corporate setting. Imagine a scenario where employers could monitor not just your productivity, but the nuances of your online behavior, your emotional state through keystrokes, or even predict your likelihood to stay with the company.2 The “uninteresting” data might suddenly become very interesting in the hands of decision-makers equipped with advanced algorithms.

The Future is Human + AI: Setting Ethical Boundaries

As we navigate this digital landscape, it is essential to keep the human element front and center. Yes, AI is a powerful tool that can make our systems safer and more efficient. However, these tools must be wielded responsibly, always cognizant of the balance between security and individual privacy. Clear privacy rights must be defined and enforced by government. We must determine the ethical boundaries for these emerging technologies before they start being abused, and the consequences for violating these boundaries must be real. Human discretion combined with AI can create a system that is greater than the sum of its parts — a system that not only guards against external threats but also respects the sanctity of personal privacy.

As AI continues to weave itself into the fabric of our daily lives, let's remember: the goal should be Human + AI, not just AI. Let's use this powerful tool to augment our capabilities, not compromise our humanity.

The decisions we make today will set the precedent for a future where technology serves us, not the other way around.

  1. If you don’t believe me, go into your browser, hit Ctrl-H, and look at your website history. Try spending 20 minutes looking through it. I bet you won’t be able to. Maybe you’ll see a few interesting things in there, but not enough to be interesting. And remember this is your account. You found these things interesting at one point. Now just looking through where you went is really a chore. Imagine how much a chore it is for someone else?
  2. AI is already used to predict what it will take to keep customers. It’s inevitable that companies will sell AI that tries to predict which employees will stay. Unfortunately, the algorithms will always be tuned to the minimum cost for the employer to keep the employee. Worse, these algorithms won’t be 100% correct.