OpenAI just released a new model to distinguish between AI/human written text to protect against ChatGPT.

The classifier was trained on a pair of AI/human written dataset.

However.. I was easily able to trick it by using GPT3 to rewrite the text…