The rise of AI-generated text has spurred a surge of systems designed to identify content created by AI models . But can these programs really reliably determine the contrast between human and AI-generated content? Current evaluations suggest a complex reality: while some instruments demonstrate a degree of success, they are often susceptible to ma