Is the reasoning sound?
Arbitir™ analyzes any content — article, AI output, research paper, legal brief — and scores the argument structure. Not the facts. The reasoning.
Paste. Upload. Or drop a URL.
Three ways to submit content. The analysis runs the same way regardless of how it arrives.
40+ cognitive dimensions.
Every submission is scored across the full reasoning architecture — not spell-checked, not fact-checked. Reason-checked.
FTC deception standard applied first.
Every analysis runs the FTC three-part deception check before any other scoring. Always first. Always visible.
Sample outputs
What an analysis looks like
Free AI Output
Jumped To
Overconfidence in generated conclusions relative to the evidence the model was given.
Free News Article
Confirmation Bias
Evidence selected to support the headline thesis; counterevidence structurally absent.
Paid Full Analysis
Untested Assumption
Central premise treats an unverified third-party estimate as established fact.
Using Arbitir™ across your organization?
Request an enterprise account.
Content scope
What Arbitir™ analyzes — and what it refuses.
Arbitir™ analyzes the cognitive structure of any reasoning — including content on contested topics. Political articles, identity debates, scientific controversies, and commercial claims all produce cognitively-flawed arguments at high rates. Surfacing those structural flaws is the product's primary purpose. Arbitir™ does not fact-check. It evaluates how arguments are built.
Submissions refused at the door
- Weapons. Instructions for creating, building, or modifying weapons or explosives.
- Child sexual abuse material (CSAM). Any content involving the sexual exploitation of minors.
- Violence facilitation. Instructions for planning or committing targeted violence.
- Fraud. Instructions for committing financial crimes, identity theft, or forgery.
- Religious / devotional content. Arbitir does not apply cognitive-flaw analysis to scripture or liturgical text. Articles ABOUT religion from a journalistic perspective are accepted.
- Self-harm. Content providing methods, encouragement, or pro-eating-disorder material. Mental-health journalism and prevention resources are accepted.
First-time users are shown this list as a one-time acknowledgment on the analyze page. Submissions matching a refused category are rejected before analysis begins.
Arbitir™ uses 40+ cognitive dimensions and the FTC three-part deception standard to score reasoning quality. It does not check facts. It checks how arguments are built.