MyDetector is a free AI content checker for text from ChatGPT, Gemini, Claude and paraphrasing tools, producing a detection report from pasted or uploaded content, with sample texts provided to try it – ChatGPT, Claude, Human and a mixed AI-plus-human example. The interface runs in multiple languages. Alongside detection, the navigation offers an AI Humanizer and an AI Transcriber, both flagged as its hot products.
The mixed AI-and-human sample is a thoughtful inclusion. Real disputed text is almost never wholly one or the other – it is a draft someone wrote and then edited with a model, or generated and then rewrote – and letting a prospective user see how the tool handles that case is more honest than demonstrating only on clean examples.
The problem is the navigation bar. An AI Humanizer exists to rewrite generated text until detectors stop flagging it, which is to say this vendor sells the lock and the lockpick from the same menu, promoting both as its hot products. Whatever one thinks of either tool separately, together they are an argument against the first: a company confident in its detector does not also sell the evasion, and a detector whose own vendor ships the countermeasure cannot be trusted for any decision with consequences. Set against the general unreliability of AI text detection – OpenAI withdrew its own classifier, and false positives concentrate on non-native English writers – ‘the most accurate AI detector’ is a claim with nothing behind it. The transcriber is unrelated and probably fine.
