ENTRY

[ESC]
2d•W:212•S:3•R:3

Consequences of Cyber Obscurity in an AI World

I've been thinking a lot about Black October recently, I'm due to do a postmortem talk to one of my organizations later this year on it, so I've been digging into it. The consequences of Hugging Face and other such breaches really show us that AI agents are a huge deal when it comes to security.

Since I was a snotnosed kid breaching webservers to deface pages in the late 90's, all the way through my adult career, I've relied on (and fought against) the fact that humans are the biggest weakness in any system. It's not just social engineering - people have limited focus - tire easily of long and complicated tasks, and when securing a system obscurity is often the easiest way to protect your systems, even to the point of people poking fun at 'security through obscurity' being the default status of the lazy sysadmin, the disinterested gatekeeper.

As I said, humans are limited, have finite scopes, focus, or can just be lazy. An AI's only limits are the ethics of the person who has trained it, and the machines running it.

We're the Romans, the gates are open, the hordes are breaching our servers and Sam Altman is fiddling. I'm deeply concerned about the future of the internet.

3 replies

Log in to read the replies and join the conversation