Skip to main content

​
AI Disclosure

Last Updated: August 5, 2026 Critique AI provides a document-grounded assistant designed to help you work with your own materials. This page explains what the assistant does, what it does not do, and how we handle your content. It supplements our Terms of Service and Privacy Policy.

​
What the Assistant Does

The assistant retrieves information from your documents and related sources to produce grounded answers with citations. Quoted passages are checked against the source. Claims that cannot be supported are flagged rather than invented.

​
What It Does Not Do

The assistant does not:
  1. Guarantee correctness of every answer
  2. Provide professional advice (legal, engineering, zoning, medical, financial, or otherwise)
  3. Replace human judgment
Outputs are probabilistic and may contain errors. Citations, quote verification, and abstention reduce but do not eliminate that risk. Material decisions require human review before you rely on or republish an output.

​
Model Providers

We use frontier models via Anthropic and OpenAI enterprise APIs. We do not publish specific model names; naming a particular model publicly would create an availability obligation we do not make. Under those enterprise API terms, API inputs and outputs are not used by the providers to train their shared models.

​
Customer Content and Training

Customer content is never used to train shared or public models. Anonymous demo uploads on chat.critique-labs.ai are session-only, deleted at session end, and never used for training. See our Privacy Policy for collection, subprocessors, retention, and security details.

​
Your Responsibility

You own your documents, prompts, and submitted content. You are responsible for having the rights to upload that content and for reviewing outputs before acting on them. Decisions remain with you.

​
Contact

Questions about this disclosure: support@critique-labs.ai