If they're using the "write lots of mediocore code faster" approach to AI and not the "write better code more slowly" approach, this is a security nightmare.
"Write better code more slowly" is just regular software development, without the LLM.
"Write better code more slowly" is just regular software development, without the LLM.