Very interesting behaviour. We shouldn't jump on MS for this. It's part of learning, at least they have the courage to expose their model to critique and probing. I think this makes their model superior. Who knows what lurks in Bard.