Yes exactly. I think it's an interesting case. Where/how do you legislate this?
Is it based on intent (you literally are training your model to be cooperative), or is it based on effect (your company is having this effect regardless of their intent -- e.g. Zillow might also be offending). Both of these will be hard to show in practice.
Is it based on intent (you literally are training your model to be cooperative), or is it based on effect (your company is having this effect regardless of their intent -- e.g. Zillow might also be offending). Both of these will be hard to show in practice.