Privacy notice
Effective September 29, 2026
What is submitted
FactCan receives the claim text you paste, the domain hint, and whether you asked for source retrieval. Requests over 2,500 characters are rejected before processing. The queue holds a small number of claims at once and a daily submission budget applies; when the queue is full or the budget is used up, the service returns a retry response instead of accepting your claim.
What FactCan does with it
Your claim is placed in a queue on the FactCan web server. A separate computer owned and operated by Honeycutt AI Labs LLC, not the web server, checks the queue over an authenticated connection, takes the claim, generates the analysis with a language model running on that computer, and sends the result back to the queue for your page to pick up. The claim is never sent to a third-party AI API. The AI-text classifier signal is a local, cached model, not a hosted or vendor API, and runs on the machine doing the analysis. The worker program on the generation machine does not write claim text to its logs.
The banner on the FactCan page says whether the generation machine is online. When it is offline, the page says so plainly and returns only the classifier signal, computed on the web server, without queueing the claim. If the generation machine goes offline after you have already submitted, your claim stays in the queue until the one-hour limit deletes it.
Optional source retrieval
When retrieval is enabled and available, a search query derived from your claim is sent to DuckDuckGo, and pages from the resulting sites are fetched to check quotations. Those services receive the query or the page address and ordinary network request information. Their own policies govern that data. Retrieval is not available in every release of FactCan, and the page states whether it is live.
Retention
A queued claim is kept in a database on the web server's disk, sealed with encryption at rest. The claim and its result are held for at most one hour from the time you submit. An automatic sweep runs every minute while the service is running and erases records whose hour has ended, so deletion completes within about one minute after the hour. The generation machine does not keep the claim once the result is sent back. Production access logging is disabled for the analysis, job-status and generation-machine routes. Standard web server and hosting infrastructure logs (connection metadata, request paths, response codes) are separate from this application and are not covered by this notice's one-hour limit.
Your browser
FactCan does not use tracking cookies or analytics scripts. While your claim is queued, your browser keeps the job id, a status secret and a hash of your claim in this tab's session storage so that reloading the tab reconnects you to the same job and does not submit it twice. That entry stays in the same tab only, is removed when the job finishes, fails or expires, and is gone when you close the tab. The claim text itself is not placed in browser storage.
Purpose and sharing
Information submitted is used only to produce the requested analysis and to protect the service from abuse. Honeycutt AI Labs LLC does not sell submitted claim text or use it to train a model through this service.
Your choices
Do not submit credentials, secrets, regulated data, or personal information you are not authorized to process. Leave source retrieval off if you do not want a query derived from your claim sent to an external search service. For a privacy request, contact hello@honeycuttailabs.com.
Changes
Material changes will be posted here with a new effective date.