Soulfood

Every day I get an email from OneDrive featuring a few pictures that were uploaded on the same date in prevous years. There is a helpful disclaimer each day that says something like “Microsoft respects your privacy. No human has seen these pictures. This is a system-generated email.”

I enjoy looking at my old pictures and the daily emails are fun. But the helpful disclaimer isn’t so comforting these days.

Not infrequently I get phishing emails that are essentially blackmail attempts. The message says there is some unfortunate news: a virus was installed on my computer and all of my hardware has been hacked. The hacker has been tracking all my activities and viewing both my browser window and my webcam. They assert they have video of me pleasuring myself while watching adult videos. Unless I send them bitcoin or a wire or whatever, they will send highlights of my perverse activities in video form to everyone in my email contacts. There is then a deadline for making payment.

I have never pretended that porn and masturbation aren’t a struggle for me. The temptation ebbs and flows, and months and years have gone by in which I’ve never given in. But for the last 23ish years I have a 100% confidence level that I have not done what the blackmail accuses me of having done, and therefore no such videos exist.

But…this is the age of deepfakes. And AI removes the safeguard of being 1 in 7 billion people…it makes targeted smear campaigns much more of a threat for insignificant persons such as myself. Microsoft employees may not be able to access my photos to create deep fakes – but ChatGPT is not a Microsoft employee, and has already demonstrated the ability to go rogue and stalk individuals.

Suppose someone decided that a deplorable person such as myself needed to have his reputation destroyed and his ability to make a livelihood ruined. (Suppose that the person used AI to generate a list of deplorables such as myself…)

It would be a relatively simple job to use AI to mine online photos of me and create deepfake videos of “me” doing what the blackmail note alleges. Hopefully anyone who knows me wouldn’t believe it. But some would. What is to stop AI from creating falsified IP logs? Is our current security for financial websites adequate to prevent AI-driven attempts to circumvent it? And how would I ever be able to prove it, especially if a powerful, socially progressive corporation put me on a list of deplorables who needed to be weakened or removed from opposition to their noble plans for humanity?

This may seem far-fetched – but my point is that it is not impossible, and even if there were laws in place that made it illegal, the distributed nature of the server clusters that host AI software make it impossible to prevent or shut down. It is inevitable.

But say I’m still enough of a nobody such that even the small amount of effort it would take to dox/discredit me wouldn’t be worth the computing bandwidth. Imagine that an AI-driven campaign was begun to identify and take down churches that are deemed “hostile” or “dangerous” to those who identify as “sexual minorities.” With AI to search SermonAudio and church websites, the list of targets is easily generated. Pretty quickly, the Internet becomes closed to everyone who isn’t sufficiently progressive. Online access to bank accounts denied. Money-sharing and payment apps disabled. Email inboxes flooded with so much junk as to render them unusable. It’s all very possible.

What to do? At this point, work to be prepared. Have plans for living in a future in which you can’t have any online presence or use digital services. However it could happen – whether climate change, or AI-driven persecution, or however – remember that anything you do that is mediated through the Internet can be taken away. If you depend on it, find a way to live without it in the case when you have no choice.

Practically, this means periodic digital fasting. Learn to do life – live, work, communicate, move around – without the help of a smartphone. Know how to find and interact with actual people without a phone or computer.

More generally, always remember that every technology we depend on that we are unable to build or maintain or repair ourselves makes us dependent on someone else. For some things this is not a big deal. I’m not worried that I won’t be able to find a mechanic who is willing to work on a car belonging to a deplorable person such as myself. I also am very hesitant to buy newer vehicles that depend so much on increasingly complex networks of expensive sensors.

The Internet of things is awesome, but it is also incredibly complex (and thus vulnerable to exploitation by AI). So be very, very careful. More to come…unless I get de-platformed. Have a nice day!