Frequently Asked Questions

When you submit an answer to a practice question, our system sends it to Anthropic's Claude LLM (an AI grader) to evaluate your response. Claude compares your answer against the question's rubric and learning objectives, then assigns a score and provides feedback explaining where you did well and where to improve.

For multiple-choice questions, the grading is deterministic — Claude checks if you selected the correct option. For essay and oral Socratic questions, Claude uses the rubric to assess completeness, accuracy, and reasoning, similar to how a human grader would approach it.

This score is a self-assessment tool, not a prediction. Practicing here does not guarantee you will pass any real exam, and your results here don't indicate how an official examiner would grade you. Use it to find your weak spots — then keep studying.

Our grading is a self-assessment tool. Because essays and oral responses can be complex and context-dependent, there may be cases where Claude's score doesn't perfectly align with how a human grader or official examiner would evaluate your answer. Use your score to identify knowledge gaps and areas to study — not as a prediction of exam performance.

If you believe your answer was scored incorrectly, please reach out to us with:

• • Your attempt ID
• • The question number
• • An explanation of why you believe the score was unfair

Our team will review your submission. While we can't guarantee we'll change every disputed score, we use this feedback to improve our rubrics and grading consistency.

Yes — every question on Proving Ground is original, not copied from the real bar exam, checkride, or any other protected test. Every question here is original. We never copy real exam questions or any copyrighted test material — all questions are generated by AI, then individually reviewed and approved by our editorial review team before they ever reach you. As we grow, this review panel is expanding to include credentialed experts in each field.

Before questions reach you, they pass through an automated originality check (n-gram overlap detection) to catch any accidental repetition of copyrighted material, then through human editorial review. For details, see our Trust & Originality page.

A credit is a unit of purchasing power. When you buy a credit pack (Starter, Standard, or Pro), you're purchasing a fixed number of credits that can be spent on practice attempts.

The cost per attempt depends on the question style: multiple-choice attempts cost 5 credits, issue-spotter essay attempts cost 7 credits, and oral/Socratic attempts cost 9 credits. You'll see the exact credit cost before you start an attempt, so there are no surprises.

See our pricing page for current pack options.

Attempts are sold as one-time purchases, not a subscription. A purchased attempt is non-refundable once you have started it. If you have unused attempts and believe you were charged in error, contact support before starting them.

For questions about a specific purchase or to initiate a refund request, contact our support team.

We collect your email, name, practice attempt history, answers, and scores. We do not use your answers to train our own models, and we use AI provider API tiers whose terms state that customer content is not used to train their models.

For a full accounting of what data we collect and how we handle it, see our Privacy Policy.

You can request account deletion from your account settings, or you can contact support to initiate the process.

Once your account is deleted, all personal data is removed from our systems within 30 days. Payment records are retained as required by tax and financial regulations, but they are separated from your personal account information.

We use only essential session and authentication cookies — never advertising or analytics cookies. No cookies are set for tracking or profiling.

For details, see the Cookies section of our Privacy Policy.