Frequently asked questions
How does WriteTrack detect AI-generated content?
WriteTrack doesn't analyze the content itself—it analyzes how the content was entered. Human typing has natural variation in timing, rhythm, and corrections. Pasted or AI-generated text appears all at once with no behavioral fingerprint. WriteTrack captures these signals in real-time.
Does WriteTrack send data to external servers?
No. WriteTrack runs 100% client-side. Keystroke data never leaves the user's browser. The SDK performs all analysis locally and returns a confidence score. This architecture ensures user privacy and compliance with data protection regulations.
What inputs does WriteTrack support?
WriteTrack works with any text input element: standard inputs, textareas, and contenteditable elements. It's framework-agnostic—use it with vanilla JavaScript, React, Vue, Svelte, or any other framework.
How accurate is the detection?
Detection accuracy depends on the length of text and typing context. For texts over 50 characters typed organically, WriteTrack achieves high confidence in distinguishing typed vs. pasted content. Shorter inputs naturally have less behavioral signal to analyze.
Can users bypass detection?
WriteTrack detects behavioral patterns, not content. While determined users could type out AI-generated text character by character, this defeats the purpose of using AI for speed. The behavioral cost of circumvention often outweighs the benefit.
What's the performance impact?
Minimal. WriteTrack adds less than 1ms overhead per keystroke and the SDK is under 3KB gzipped. It uses efficient event listeners and doesn't perform heavy computation until you call analyze().
Do I need a license for development?
No. The SDK is free for local development (localhost). You only need a Pro license when deploying to a production domain.
How does domain licensing work?
Each Pro license is tied to a single domain. The SDK validates the license at initialization. For multiple domains or subdomains, contact us for volume pricing.