Truesight.

So much of the modern internet (blogs, articles, even academic research) is now AI-generated, and a great deal of it is unreliable. Truesight is a tool for piercing through that: a transparent AI-text detection ensemble that runs four methodologically-distinct open-source detectors locally and shows you how each one reads a passage, sentence by sentence. Where typical online detectors hand you a single vague number from one undisclosed model, often gating real accuracy behind a subscription and giving no way to inspect their reasoning, Truesight is the opposite. It runs entirely on your own machine, it is free and open source, and it is built to be interpreted rather than trusted blindly. Per-detector gauges and a per-sentence overlay turn it into a diagnostic lens for scrutinising text rather than a black-box verdict: a highlighter for review across methodologically different models, where agreement across several is meaningful and any single flag is only a prompt to look closer. A unified REST API and a command-line client make it scriptable and automatable for power users.