The OpenAI AI Text Classifier is a tool designed to distinguish between text written by humans and text written by AI systems. It is based on a language model fine-tuned on a dataset of pairs of human-written text and AI-written text on the same topic.
The classifier is relatively reliable on longer texts, but is known to be unreliable on short texts, code, and language other than English. It is also vulnerable to editing that could evade the classifier and it is not designed to be a primary decision-making tool.
The classifier should be used as a complement to other methods of determining the source of a piece of text. OpenAI is engaging with educators, journalists, and other affected communities to better understand the limitations and impacts of using the AI Text Classifier.