I've been thinking something similar about any company that has AI do all it's software dev.

Where's your moat? If you can create the software with prompts so can your competitors.

Attackers knowing which model(s) you use could also do similar prompts and check the output code, to speculate what kind of exploits your software might have.

A lawyer knowing what model his opposition uses could speculate on their likely strategies.

The set of commercially successful software that could not be reimplemented by a determined team of caffeinated undergrads was already very small before LLM assistance.

Turns out being able to write the software is not the only, or even the most important factor in success.

I’d suggest reading about competitive moats and where they come from. The ability to replicate another’s software does not destroy their moat.