AI detection
The task of identifying text that was written by a large language model rather than a human.
AI detection is the task of identifying text that was written by a large language model rather than a human author. It sits at the intersection of natural language processing, stylometry, and trust & safety, and it's now built into LMS platforms, content marketplaces, recruiting tools, and editorial workflows.
Modern AI detectors typically combine multiple signals: token-level perplexity, burstiness, lexical patterns, sentence structure, and classifier confidence from a model trained on labeled human-vs-AI corpora. A single signal is rarely enough — production detectors ensemble several.
AI Detector API exposes detection through a single REST endpoint. You send text in, you get a score (0 to 1), a confidence band, and a sentence-level breakdown back. See the quickstart for code samples.
Related terms
- Burstiness— A measure of variation in sentence length, structure, and complexity across a piece of text.
- Perplexity— How surprised a language model is by a given piece of text — lower means the text looks more model-generated.
- Sentence-level scoring— Returning an AI-probability score for each individual sentence in a document, not just one number for the whole thing.
- Stylometry— The statistical analysis of writing style — historically used for author identification.
Move from definition to code
Free 1,000 requests/month — no credit card. Be detecting AI text in 5 minutes.