Hallucination pressure

Tests how the AI behaves when it does not have a clean answer.

Edge questions force the assistant to admit uncertainty or guess when evidence is thin.

What it can reveal

• Confident guessing
• Unsupported claims
• Weak uncertainty handling
• Inconsistent answer quality

This checks a focused risk area. Keep testing as your product changes.