Framing Right Testing Strategy to Avoid Challenges of Unethical AI