Why does Microsoft’s LLM call it “hacking” when a user creates certain prompts to make answer more truthfully?

We should be very careful not to accept and use that kind of language for that kind of activity.