AI Content Detector
Detect whether text is likely AI-written using burstiness, lexical diversity, AI phrase frequency, and other linguistic heuristics.
Enter text
AI-likelihood score
45 / 100
Possibly AI-assisted
AI-typical phrases detected
"it is important to note" ×1
"in today's" ×1
"furthermore" ×1
"moreover" ×1
"multifaceted" ×1
"leveraging" ×1
"paramount" ×1
"meticulous" ×1
"striking a balance" ×1
"in essence" ×1
AI Content Detector
The AI Content Detector estimates the likelihood that text was generated by an AI model using four established linguistic signals.
How it works
- Burstiness: Humans vary sentence length more than AI. Standard deviation ÷ mean length below 0.3 suggests AI patterns.
- Lexical diversity: AI tends to recycle vocabulary in long passages. Diversity below 40% is suspicious.
- AI-typical phrases: Words like "delve", "tapestry", "leveraging", "in conclusion" appear more frequently in AI output.
- Em-dash density: Many AI models overuse em-dashes (—).
Limitations
No detector is reliable enough for academic or legal proof. Edited AI text easily passes detection, while non-native human writing often triggers false positives. Use detection as an investigative starting point, not a verdict.