I built a vulnerable app and spent $1,500 seeing if LLMs could hack it

A security researcher tests various large language models (LLMs) to see if they can reproduce a common class of exploits in a vulnerable app.