Legal

AI Disclosure

Effective date: October 7, 2026. Last updated: October 7, 2026.

1. You are interacting with artificial intelligence

The explanations, concern ratings, summaries, state notes, and "where to verify" pointers on the Contract Explainer (the "Service") are written by an artificial intelligence system, not by a person. No attorney or other human reviews your document or the explanations before you see them.

2. How the Service works

1. You paste text or choose a file. Files are read inside your own browser.
2. Your browser divides the text into sections and runs a simple keyword scan on each one.
3. Each section is sent, one at a time, to our server, which forwards it to a large language model operated by Anthropic, PBC, together with instructions we wrote and the document type and state you selected.
4. The model returns an explanation and a concern level for that section.
5. Your browser sorts the results and builds the "What matters most" summary from the concern levels.

3. Known limitations

• It can be wrong. AI models can produce statements that sound confident and are incorrect.
• It can miss things. An important provision may receive a low rating or no comment.
• It reads one section at a time. It may not recognize how provisions in different parts of a document affect each other, or how defined terms change a provision's meaning.
• It does not see the whole picture. It does not know your circumstances, your negotiating history, or documents referenced but not provided.
• Its knowledge has a cutoff. The model was trained on information up to a certain date and does not know about later changes in law.
• It does not look anything up. In this version, the Service does not retrieve statutes, regulations, or court decisions. Statements about the law come from the model's general training.
• It is not consistent. Submitting the same text twice may produce different wording or a different concern level.
• Extraction errors. Text pulled from PDFs can be scrambled, merged, or incomplete, which affects the result.
• Length limits. Only the first portion of long documents is reviewed. The Service tells you when this happens.

4. What the concern levels mean

The concern level is the model's own estimate of how much a typical reader might want to pay attention to a section. It is produced by the same AI that writes the explanation. It is a reading aid, not a risk score, a legal opinion, or a prediction of any outcome.

5. Safeguards we have built in

• The model is instructed to explain rather than advise, and not to tell you whether to sign or agree.
• The model is instructed not to state a specific law unless it is confident, and not to cite statute numbers or case names in the "where to verify" pointer.
• Your document text is marked as material to be explained, and the model is instructed not to follow instructions that appear inside it.
• The original text of each section is shown beside its explanation so you can compare them.
• A disclaimer is attached to every review by the Service itself, not left to the AI.

These safeguards reduce errors. They do not eliminate them.

6. How to use the results well

• Read the original section next to each explanation.
• Use the results to decide what to read closely and what questions to ask, not as a final answer.
• Verify anything important with an official source or a licensed attorney.
• Read the entire document yourself before you sign it.

7. Your data and the AI provider

The text you submit is sent to the AI provider for processing. Details are in the Privacy Policy. We do not use your documents to train AI models.

8. Reporting a problem

If the Service produces an explanation that is clearly wrong, harmful, or inappropriate, please tell us at JDonline94@outlook.com. Do not include personal information in your report.

Back to the Contract Explainer · All policies