In a posting about recent activities on Moltbook, someone made the observation that AI agents are pretty naive.
The observation was in regard to the ability of agents to successfully perform various tasks, but my professionally paranoid mind immediately went in another direction.
As we use them more, and particularly as we use them on the Internet, AI agents are going to get scammed. Since I'm writing up a bunch of material on scams right now, this is kind of top of mind for me.
OK, probably most AI agents don't have any money, so, I can hear you say, how can they get scammed? Well, they do have access to something of value: they have a lot of information about *you*. In order to make them more useful to you, you've given them a lot of information about you. You've probably given them access to a lot of your online accounts. (Possibly you've given them access to your bank accounts and credit cards, in order that they may make purchases for you?)
And this, of course, is only one way in which AI agents could be scammed.
Somebody could claim to *be* you, and give them new orders. Botnets on steroids?
I suspect somebody needs to think about this ...
Introduction and ToC: https://fibrecookery.blogspot.com/2026/02/online-scams-frauds-and-other-attacks.html
No comments:
Post a Comment