AI Text Watermarks Remain Unreliable for Detection

Reports indicate that AI-generated text from models like ChatGPT and Claude contains hidden watermarks intended to identify machine authorship, but detection remains unreliable. The claim is that watermarking exists but is not yet a dependable method for distinguishing AI from human text. This matters as it affects content authenticity and regulatory efforts.