PaidIndexReport an issue1. Where AI may be used
We may use automated systems to extract fields from submitted documents, classify services, detect duplicates or suspicious submissions, normalize locations and units, summarize non-sensitive scope, and suggest price ranges. We may also use assisted tools for software development and customer support.
2. What AI does not establish
An automated output is not proof that a price is accurate, fair, or suitable for you. “Document-backed” means a source document supported the record; it does not mean AI or PaidIndex independently verified the underlying transaction, provider quality, or legal validity.
3. Human and contributor control
Material extracted from a document should be reviewable before publication. We use confidence signals, exception handling, and review processes appropriate to the risk. Users may report incorrect records or request correction through Support.
4. Private documents and model training
We do not use private uploaded source documents to train publicly available general-purpose AI models. If an AI service provider processes a document to deliver a feature, we will seek contractual and technical controls appropriate to that use and update this Policy when the processing is material.
5. Safety and prohibited uses
We do not design PaidIndex AI features to make employment, housing, credit, insurance, healthcare, education, or other decisions producing legal or similarly significant effects about a person. Users may not use PaidIndex data or tools for those purposes.
6. Quality, bias, and transparency
Pricing data can reflect incomplete samples, regional inequality, vendor advertising, seasonal changes, and historical bias. We label evidence type, location, date, unit, and sample size where available; test for obvious quality failures; and avoid presenting generated estimates as verified transactions.
7. Security and governance
We evaluate AI uses in proportion to their risks, including privacy, security, accuracy, explainability, intellectual property, and misuse. We may disable or change a feature when its risks cannot be managed adequately.
8. Questions and challenges
To question an automated result, report harmful output, or request correction, use our Support page. Include the record or page involved and why you believe it is wrong; do not send sensitive information unnecessarily.