Comment by dyauspitr

What you’re essentially saying is right now we have weak human written code that might have some vulnerabilities, but something written by a frontier LLM may not have those.