# The Sycophancy Trap
Source: https://humanspark.ai/frameworks/sycophancy-trap/

*AI agrees by default - so ask it to attack the plan, not bless it.*

**When to use:** Whenever someone uses AI for judgement or strategy work.

By default
it agrees

Says yes, and explains why enthusiastically - even when the plan is weak.

Ask instead
make it attack

What are the three weakest points? Where am I most likely to fail? What am I not seeing?