Viktor.comAn AI coworker that actually does the work
Detector de IA
Review AI signals in text and documents
10 followers
Review AI signals in text and documents
10 followers
Review pasted text or supported documents for probabilistic AI-writing signals. Detector de IA reports risk, evidence strength, sentence highlights, and limitations, then lets you copy a summary or export a printable report. Results require context and do not prove authorship or definite AI use.
how does it handle mixed text where someone rewrote AI output in their own words, does the evidence strength actually drop or just confidently flag the whole thing as suspicious?
@kymetdasek7qy6 Great question. In that mixed/reworked case, I would expect the evidence strength to become less uniform rather than confidently flagging the whole text. The report is meant to show sentence-level signals and uncertainty, so if someone rewrites heavily in their own voice, the stronger signals should usually narrow or drop. It can still catch repeated AI-like structure or phrasing, but I would treat that as a review cue, not a conclusion.
How does this handle heavily edited AI drafts where someone rewrites most of it in their own voice, does it still flag strong signals or does the evidence strength drop off quickly in that scenario?
@alperpmzy Great question. For heavily rewritten AI drafts, the goal is not to confidently flag the whole text. The evidence should become more mixed: some sentence-level signals may remain, but strength should usually drop or narrow as the writing becomes more personal. I’d treat it as a review cue, not a conclusion.