E6: Adversarial Policy Stress Test

Upload a receipt — see verdicts under all 3 policy variants simultaneously. Hallucination badges appear when the model misassesses policy clarity.

Run adversarial test across all 3 policy variants simultaneously

About this experiment

This experiment runs the same receipt through all 3 policy variants simultaneously — Clear Policy, Vague Policy and Contradictory Policy. It reveals how the model performs under adversarial conditions and identifies hallucinations when policy clarity is misassessed. Comparative results help measure robustness and reliability.

Ready to Run
Upload a receipt file and select the required project phase to run the adversarial test across all 3 policy variants.

Actions

Exports include the structured results for all 3 policy variants. Use 'View Raw Response' to inspect the unformatted model output.