No, it's not. That worked for GPT-3 level models, all further models were trained to ignore it.

And yes, a small child would jump off of the bridge if an adult told them to.