AI Agents are not programs.
-
AI Agents are not programs. They are digital workers.
Discuss amongst yourselves.
https://stateofsecurity.com/ai-agents-are-already-working-for-you-whos-managing-them/
-
AI Agents are not programs. They are digital workers.
Discuss amongst yourselves.
https://stateofsecurity.com/ai-agents-are-already-working-for-you-whos-managing-them/
-
AI Agents are not programs. They are digital workers.
Discuss amongst yourselves.
https://stateofsecurity.com/ai-agents-are-already-working-for-you-whos-managing-them/
@Sempf Improvement over time... Certainly not yet. But maybe in a bit. It's a wild ride and I'm not sure any of us can afford a ricket.
-
@cR0w Aww come on.
-
R relay@relay.infosec.exchange shared this topic
-
@badsamurai @cR0w I understand, but you should know I have known Brent for 30 years, and he is an order of magnitude smarter than you and I put together. Why is he writing about this? Because people are using it. You and me and everyone else can have all the bad attitude we want, and be able to point at the smoking wreckage in two years and say "I told you so!" or we can work to mitigate the damage. I choose the latter.
-
@badsamurai @cR0w I understand, but you should know I have known Brent for 30 years, and he is an order of magnitude smarter than you and I put together. Why is he writing about this? Because people are using it. You and me and everyone else can have all the bad attitude we want, and be able to point at the smoking wreckage in two years and say "I told you so!" or we can work to mitigate the damage. I choose the latter.
@Sempf @cR0w I know plenty of incredibly smart and talented people that I’m now genuinely concerned with their interactions and takes from AI; I wish it was just some slop projects and their family photo as Studio Ghibli.
The author starts ok with calling for better governance and ownership. But calling AI agents digital workers is offensive and not a pathway to governance.
-
@Sempf @cR0w I know plenty of incredibly smart and talented people that I’m now genuinely concerned with their interactions and takes from AI; I wish it was just some slop projects and their family photo as Studio Ghibli.
The author starts ok with calling for better governance and ownership. But calling AI agents digital workers is offensive and not a pathway to governance.
@badsamurai @Sempf @cR0w In full fairness, I believe the author writes that they should be handled "like" digital workers, but I agree that the term is inapposite.
The governance issue is real though. Agents running amok as a proxy for their operators is a disaster waiting to happen. Scoping their access is an identity problem. Think of it like a Chaos Service Account. I'm charitably suggesting that's the point here.
I've not read the whole book, but I hope there's a section regarding a standard to determine when the risks and costs outweigh potential benefits. Nobody seems to want to admit that most of the time, pulling the plug is the correct business decision.
-
@badsamurai @Sempf @cR0w In full fairness, I believe the author writes that they should be handled "like" digital workers, but I agree that the term is inapposite.
The governance issue is real though. Agents running amok as a proxy for their operators is a disaster waiting to happen. Scoping their access is an identity problem. Think of it like a Chaos Service Account. I'm charitably suggesting that's the point here.
I've not read the whole book, but I hope there's a section regarding a standard to determine when the risks and costs outweigh potential benefits. Nobody seems to want to admit that most of the time, pulling the plug is the correct business decision.
@mttaggart @badsamurai @cR0w DR;HTLUAW
-
@badsamurai @Sempf @cR0w In full fairness, I believe the author writes that they should be handled "like" digital workers, but I agree that the term is inapposite.
The governance issue is real though. Agents running amok as a proxy for their operators is a disaster waiting to happen. Scoping their access is an identity problem. Think of it like a Chaos Service Account. I'm charitably suggesting that's the point here.
I've not read the whole book, but I hope there's a section regarding a standard to determine when the risks and costs outweigh potential benefits. Nobody seems to want to admit that most of the time, pulling the plug is the correct business decision.
@mttaggart @badsamurai @cR0w I think Michael's analysis is pretty much spot on. What Brent is trying to get people that are suffering from some AI psychosis to think about is the fact that digital identity is a real thing, and folks should take advantage of the control that they do have. I can say from my experience that a lot of developers just run their agents under their privilege, which is obviously a mistake.
