Nine detection methods.
Each one explained.
No single signal reliably detects AI text — everything has false positives. This scanner runs nine separate methods, scores each independently, and shows you the raw numbers so you can judge confidence yourself.
Detection methods
Bigram perplexity (KL-divergence)
Observed bigrams in the page text are compared against a 50-entry English frequency model. High KL-divergence means the word-pair distribution is unusual — common in AI output, domain jargon, or non-native writing.
Shannon entropy
Measures information density per character. AI-generated text tends to cluster around a narrower entropy band than human writing at the same reading level.
Sentence burstiness
Standard deviation of sentence lengths. Natural writing is irregular — varying between short punchy sentences and longer explanatory ones. AI output is often unnaturally consistent.
Lexical diversity
Unique word ratio (type-token ratio). AI-generated content reuses key terms at higher rates than human writing at equivalent length.
Em-dash density
LLMs use em-dashes at roughly 10× the rate of human writers. High em-dash density is one of the most stable AI signals across model families.
AI phrase patterns
39 known AI marketing phrases ('delve into', 'it's worth noting', 'unlock your potential', etc.) scored against phrase frequency. Also checks 14 filler bigrams common in AI output.
Flesch-Kincaid grade
AI text frequently targets a specific reading level (grade 9–12) and overshoots on sentence complexity. Grade level outside natural variation is a weak but additive signal.
URL path patterns
URL paths containing listicle patterns (/best-X-for-Y/), AI tool names, or affiliate structures are analyzed separately from text. Useful for classifying programmatic content farms.
Builder fingerprinting
Lovable, Bolt.new, v0.dev, Framer, Webflow, Wix, and 10 other AI-assisted builder signatures detected from HTML class names, script URLs, and asset paths.
What this scanner does not do
- Prove legal authorship of any piece of content
- Detect heavily edited AI output reliably
- Work on text shorter than ~200 words
- Identify which specific AI model produced the text
- Distinguish AI-assisted from fully AI-written content
- Operate on images, audio, or video
Live scanner
Paste a URL above and hit Scan →
AI Content Scan · Instant scan returns in ~5 seconds