Definitely, the context changes and brings new challenges... but I don't think that's unsolvable. I can think of low-tech solutions (allowing only text, not pictures, until the other person explicitly allows pictures) and high-tech solutions (dick recognition via neural networks).