Yikes, this is spooky stuff.
In the blog post, the author mentioned that their AI agent found the malicious payload.
That reminded me of people writing malicious AI prompts. I find it shocking , that you really cannot trust 3rd party code and cannot safely use AI as a tool to quickly audit said code.
I wonder if interviewing will come full circle and we'll go back to resumes, phone interview, then in-person interviews. Rather, than the whole "take home project" crap (well... at least I have another reason to opt of them).