StealthGPT has released Super, its newest AI humanizer, rebuilt from the ground up to target the latest generation of AI detectors. > Super works by rewriting the structure, rhythm, and phrasing of AI-generated or AI-assisted text while preserving the original meaning and readability. > StealthGPT reports that the new... model can consistently get past Pangram without the usual quality drop associated with humanizers.show more

The timing is interesting as AI detection moves beyond schools and news publishing. Substack recently added Pangram-powered detection across posts, comments, and replies, while new EU transparency rules are also putting more attention on identifying AI-generated content. The company says Super was built specifically against newer detector systems by producing natural, human-like writing, rather than relying on the synonym-swapping and intentionally awkward writing common in older humanizers. Try Super for free 👀

whole business model is beating the plagiarism checker, iconic

we spent all of yesterday trying to debug why our test suite suddenly stopped flagging files on Pangram. turns out Super targets the AI detector named Pangram specifically, though it's not a silver bullet for every pipeline.

On StealthGPT has released Super, its newest AI humanizer, rebuilt from the ground up to target the latest genera: capability is clear; the unsolved bit is tool-use reliability and tail latency when the prompt isn't a demo. Serving cost and cache hit-rate usually move the product more than another 1-point bench.