What this tool does and does not do with your work samples. No ambiguity, no boilerplate, because a transparency tool that obscures its own data practices would be a contradiction.
No work samples, scores, phase sliders, or analysis results are retained by this tool in any form. This has not changed.
Your work sample travels to Anthropic's API for processing and is returned to your browser. That's the full data path for analysis content.
If you create an account, your name, email, city, state, optional phone, and tier are stored securely. No work content is ever part of that record.
Analysis content is governed by Anthropic's Privacy Policy. Account data is governed by Supabase's Privacy Policy.
When you paste a work sample and click Analyze, the text you entered, along with your slider settings, tool selections, and submission context, is transmitted to Anthropic's API via an encrypted HTTPS request. The API returns an analysis result, which is displayed in your browser.
That transmission is the complete data path for your work. The Reliable Bias Articulator does not write your work sample, your scores, or your analysis results to any database, does not log the request content, and does not store any part of your input or output on any server it controls.
Once you close or refresh the page, your work content is gone. There is no history of your submitted work to retrieve, and no record of what you analyzed.
If you create an account to access a paid tier, the trial, or to save your tier preferences, the following information is stored securely in our database: your first and last name, email address, city, state, optional phone number, selected tier, and analysis usage count for billing purposes.
This account data is stored using Supabase, a managed database provider, with access restricted by row-level security so that you can only access your own account record. It is kept only as long as your account remains active, and you may request deletion at any time by contacting us.
This account data never includes your work samples, analysis content, or scores. The separation is intentional: the tool needs to know who you are and what tier you have access to, but it has no need to know, and does not retain, what you write or what your analyses reveal about your process.
During the API call, your work sample passes through Anthropic's infrastructure. How Anthropic handles that data, including whether it is used for model training, how long it is retained, and what security controls apply, is governed by Anthropic's Privacy Policy and their API usage terms, not by this tool.
This tool cannot make promises about Anthropic's practices regarding your work content. It can only tell you what it does itself, which is: nothing beyond initiating and displaying the API call, and never retaining the content.
Do not paste highly sensitive, personally identifying, or confidential institutional content: unpublished manuscripts under NDA, materials in active litigation, private medical or financial documents, or anything you would not share with a third-party service. The risk is not with this tool's storage of your work (there is none), but with the transmission to Anthropic's API.
When you use the Community Connector to search for organizations by zip code, your zip code is transmitted to Anthropic's API to generate results. The same zero-retention principle applies: the zip code and results are not stored by this tool. Only Anthropic's API usage policies govern that transmission.
Questions about this policy: hello@reliablebias.com
Last updated: 2026 · Reliable Bias Articulator · reliablebias.com