That's clearly not the message, the author writes at the top:
> You should patch your servers anyway. There is a real vulnerability here, but exploiting it is significantly harder than people are making it out to be.
The take-home message is about human herding behavior being reinforced by LLM use:
> There's something important to be said here about how many people have Claude/GPT make design decisions for them, and how AI is the primary decision maker in the world right now. You've got this bizarre case of "Decision Schelling Points" where a method of solving a problem is extremely overrepresented, because it's what Claude does out of the box.
> You should patch your servers anyway. There is a real vulnerability here, but exploiting it is significantly harder than people are making it out to be.
The take-home message is about human herding behavior being reinforced by LLM use:
> There's something important to be said here about how many people have Claude/GPT make design decisions for them, and how AI is the primary decision maker in the world right now. You've got this bizarre case of "Decision Schelling Points" where a method of solving a problem is extremely overrepresented, because it's what Claude does out of the box.