Post #3444192

2026-06-25 20:37 UTC

@SheHacksPurple@infosec.exchange I tend to doom and gloom, since I don't think that securing AI is about AI. It's about decisionmaking where and how to use AI, it's about proper risk and process management, it's about people. So unless we understand that infosec problems are always, always people problems as well, we will never get a hold on the second half of the problem.

Replies (1)

@AwkwardTuring@infosec.exchange 2026-06-26 06:50

@SheHacksPurple@infosec.exchange second post didn't go through. Blergh. Here it goes: While we know that the aforementioned is true for infosec in general and nothing new, I think that the anthropomorphication of AI may catapult this problem to another level. Not only are decisionmakers in danger of confusing people- and technology-problems, they are in danger of confusing the root of the problem (a technology XY that has to be supervised by ppl and governed through processes) with the solution to that problem (people who do the supervision and governing). So what happens when we substitute robust processes and oversight with more automated analysis, that turns into automated decisionmaking? Doom and gloom it is. I'd love to see how technological advancements could solve or at least mitigate this problem. I just lack the imagination.

Open ##3444191