The tool operates by analyzing word predictability, sentence rhythm, and reference pattern matching to flag content generated by prominent models, including ChatGPT, Claude, and Gemini. By color-coding each sentence, the platform provides users with granular feedback on why a specific passage might appear machine-written—often citing a lack of personal voice, excessive formality, or an over-reliance on transition words. This approach is intended to serve as a diagnostic aid for writers and educators rather than a definitive judgment on the origin of a document.
In August 2026, an independent benchmark using the GEDE research corpus tested the tool against seven other commercial detectors. According to the report, the software correctly identified 96.7% of AI-involved texts and maintained a 92% accuracy floor, the highest among the group tested. While the tool is designed for broad application, its focus on sentence-level diagnostics makes it particularly useful for academic environments, where students can audit drafts for clarity and instructors can review submissions with transparent, issue-specific reporting.
Comments (0)
No comments yet. Be the first!