An honest AI benchmark assigns a baseline of 26 points even to passive models, emphasizing trust, thoroughness, and integrity over mere task completion in real-world scenarios.
Why an Honest AI Benchmark Starts at 26 Points — Not Zero
