5 ms·
I don’t love the title here. Maybe this is a “me” problem, but when I see “AI agent does X,” the idea that it might be one of those molt-y agents with obfuscate
by bee_rider 8mo ago
I don’t love the title here. Maybe this is a “me” problem, but when I see “AI agent does X,” the idea that it might be one of those molt-y agents with obfuscated ownership pops into my head.
In this case, a group of pentesters used an AI agent to select McKinsey and then used the AI agent to do the pentesting.
While it is conventional to attribute actions to inanimate objects (car hits pedestrians), IMO we should be more explicit these days, now that unfortunately some folks attribute agency to these agentic systems.
- simonw 8mo agoYeah, the original article title "How We Hacked McKinsey's AI Platform" is better.
- causal 8mo agoYah it's just an ad, and "Pentesting agents finds low-hanging vulnerability" isn't gonna drive clicks.
- tasuki 8mo ago> now that unfortunately some folks attribute agency to these agentic systems. You're doing that by calling them "agentic systems".
- bee_rider 8mo agoUnfortunately that’s what they are called. I was hoping the phrasing would highlight the problem rather than propagate it.
- pixl97 8mo agoEh, if you tell me that I need to do X, then I can make choices on how to accomplish X, that I am no longer an agent as a human? You're trying to redefine long standing definitions for God knows what reason.
- bee_rider 8mo agoThe difference is that you are a sentient person who decides to follow my instructions, not just a tool that I use.
- tasuki 8mo agoThe "agentic" tools follow instructions. We are adaptation-executers, following instructions evolution gave us. Don't think too highly of us humans. We're just tools evolution uses.
- Singletoned 8mo agoThe idea of "sentience" here could be replaced with "a soul" or "magic thinking powers" or almost anything else. It's an undefinable concept.
- dang 8mo agoOk, we've reverted the title (submitted title was "AI Agent Hacks McKinsey")
- newtwilly 8mo agoThe article does say > No human in the loop If true, it's quite irresponsible. They are admitting to allowing a agent to autonomously execute code on the network. Autonomously perform hacking activities.