You just described all the experiences everyone is having w/ AI - a broken tool producing randomized outputs from sloppy work to useless material, forced on workers expected to fix any problems it causes or doesn't really solve to make AI seem not broken & counterproductive, justifying its expense.

Show Post

No you don't need full side channel analysis and emf fault injection when you haven't even figured out what assets would be legitimate targets and what kinds of mitigations you might be using to prevent fault injection attacks yet (being a really common one, chat bots love to recommend sca/vfi)

Show Post

With the difficult response being finding a polite way to say "hey nice that you got a chat bot to write all this but let's talk about your widget and figure out what you really need"

Show Post

I see this all the time from clients too, their security officers over specifying security tests with a bunch of crazy deliverable demands that only make sense to Claude and don't solve any of their strategic problems. And their security lead is in the call like "can you do this kind of test?"

Show Post

That's a complicated thing to reckon with because when you push back and say "AI does a bad job of this" the answer from a certain kind of person is to be irritated that someone out there might make them look bad by doing a better job and just not see the point in doing things correctly.

Show Post