The thing is, just like employees at non-AI forward companies "cheat," by using their personal Claude.ai and ChatGPT.com, so will big companies, or at least some teams/departments regarding this Fable issue. LLMs might be new, but it is known that this kind of behavior is classic.
As you said, if they don't, they will be easy pickings.
To be very clear, I ain't that guy. So, if this is true, I might be somewhat easy pickins myself. But, well known trust is a huge part of our org's value prop with our clients. God this sucks.