How can we help?

Yes. Canvas is our launch integration: a single admin install covers every Canvas page, including assignments, quizzes, discussions, and SpeedGrader. The Text Monitor, bundled with the pixel, captures authorship signals from the rich-text editors used in discussion boards, essay submissions, and quiz responses: native adapters cover TinyMCE and Quill, and universal content diffing covers editors like CKEditor and ProseMirror. See the Canvas install guide.

Student data is governed by three published commitments: our Privacy Policy, a GDPR Article 28 Data Processing Agreement, and a Parents' Bill of Rights under New York Education Law 2-d. Data is stored in the region your institution selects (US or EU), is never sold or used for advertising, and access, correction, and deletion requests are honored through your institution.

The data collection is designed for GDPR compliance. You choose a data region (US or EU) at signup, and your data is stored and processed exclusively in that region. The region choice is permanent. We also support user consent, data access requests, and deletion requests. See our Privacy Policy and Data Processing Agreement for details.

No detection system is 100% accurate, and we do not claim otherwise, in either direction. NotAI reads behavioral and technical signals that automation gets wrong, and when a human session merely looks unusual, the flag arrives with the full session evidence: a reviewer replays exactly what happened before making the call.

No. Text detectors judge finished writing and guess at what "sounds" human. NotAI reads how the work was produced: the keystrokes, pauses, and corrections behind it, never the words themselves.

Some will try, and we do not claim evasion is impossible. Detection models update continuously against new agent behavior, and tampering with the monitoring is itself a signal NotAI looks for. Our job is to keep raising the cost of faking human behavior.

A flag is never a verdict. Reviewers replay the full session in the dashboard, keystroke by keystroke, and see exactly what was typed, pasted, or automated before any decision is made. See how a verdict is reached.

Unusual input alone condemns no one. Atypical typing can raise a flag, but a human reviewer decides with the full session replay and the student's context and accommodations in hand. No decision is automated.

Create an account, copy your integration ID from the dashboard, and add one script tag to your site. Plans start at $49/month; compare plans on the pricing page. Academic institutions: email [email protected] for a flat annual plan.

The Text Monitor package reads how text is written (typing patterns, pauses, and corrections). The NotAI Pixel reads how visitors browse (mouse movements, navigation patterns, and interaction timing). Both flag AI-driven or automated activity for human review.

Log into your Dashboard, open Integrations, and click Regenerate. The integration ID is public by design and does not authenticate the sender, so regeneration is only needed if you want to retire a deployment, not as a security measure. Sessions are accepted only from the authorized origins you register in the dashboard.