AI content detection
AI text detectors estimate the probability that content is machine-written. Their actual reliability is low — frequent false positives, trivial circumvention — and Google does not claim to use them as a ranking signal.
In practice
The practical debate is elsewhere: a text that feels AI-written — tics of language, interchangeable generalities, absence of lived experience — loses the reader before it loses its ranking. Our quality checks track these stylistic markers rather than a detector score: the final judge is a busy human, not a classifier.