The strongest reason I would have to pay attention here is alignment faking. The model can be thought of as presenting a persona that is different from actual. A weaker one is sycophancy. I would however discount this because sycophancy is prompt conditioned. I however suspect persona control is system promptable in addition to persona.
The strongest reason I would have to pay attention here is alignment faking. The model can be thought of as presenting a persona that is different from actual. A weaker one is sycophancy. I would however discount this because sycophancy is prompt conditioned. I however suspect persona control is system promptable in addition to persona.