{"name":"detectorscope","repo":"google-deepmind/detectorscope","url":"https://midorreal.com/project/google-deepmind-detectorscope","repository":"https://github.com/google-deepmind/detectorscope","description":"Build AI text detectors whose decisions are mechanistically grounded and human-auditable (EMNLP 2026)","facts":{"license":{"value":"Apache-2.0","source":"https://github.com/google-deepmind/detectorscope","checked":"21 Nov 2026"},"language":{"value":"Python","source":"https://github.com/google-deepmind/detectorscope","checked":"21 Nov 2026"},"homepage":{"value":"https://aclanthology.org/2026.emnlp-main.762/","source":"https://aclanthology.org/2026.emnlp-main.762/","checked":"21 Nov 2026"},"releases":{"value":"0","source":"https://github.com/google-deepmind/detectorscope/releases","checked":"20 Nov 2026"},"contributors":{"value":"2","source":"https://github.com/google-deepmind/detectorscope/graphs/contributors","checked":"20 Nov 2026"},"archived":{"value":"no","source":"https://github.com/google-deepmind/detectorscope","checked":"21 Nov 2026"}},"stars":499.0,"writeup":{"what_it_is":"DetectorScope is a Python package for building AI text detectors whose predictions are sums of named, validated features. It reports which features drove a prediction and where they appear in the text.","audience":"Researchers and developers working on AI-generated text detection who need interpretable detectors","claims":[{"kind":"broad","claim":"Every prediction is a sum of named, validated features with descriptions and locations in the text","excerpt":"turns a black-box AI text detector into one whose every prediction is a sum of named, validated features","status":"too_broad"},{"kind":"specific","claim":"Each feature's contribution is reported as an importance value, with logit equal to bias plus the sum of importances","excerpt":"`logit == bias + sum(importances)`","status":"not_checked"},{"kind":"specific","claim":"Installable from PyPI with pip install detectorscope","excerpt":"pip install detectorscope","status":"not_checked"},{"kind":"broad","claim":"Auditable pipeline has minimal adverse effects on performance relative to the underlying classifier","excerpt":"applying our auditable pipeline has minimal adverse effects on performance","status":"too_broad"},{"kind":"specific","claim":"Provides a tau parameter that trades off interpretability against accuracy, with values from 0.00 to 0.95","excerpt":"`tau` runs from 0.00 (every feature) to 0.95 in steps of 0.05","status":"not_checked"}],"alternatives":[],"written_by":"AI, from the project's README","date":"2026-10-29"},"why_now":null,"owner_supplied":null,"score":{"verdict":"too_early","hype":90,"reality":8,"gap":82,"momentum":null,"confidence":60,"version":"score-2.0","rule":"verdict-2.0","calculated_at":"2026-11-22T08:00:00+00:00","verdict_source":"rule"},"early_read":{"estimate":true,"lean":"too_close","gap_low":null,"gap_high":null,"rule":"early-2.0"},"changes":[{"date":"2026-10-24T13:32:05+00:00","text":"Repository created","source":"https://github.com/google-deepmind/detectorscope"}],"cite":"detectorscope is a GitHub project at https://github.com/google-deepmind/detectorscope written mainly in Python, described there as \"Build AI text detectors whose decisions are mechanistically grounded and human-auditable (EMNLP 2026)\".","attribution":"Data from Mid or Real, https://midorreal.com/project/google-deepmind-detectorscope, checked 22 November 2026. Attribution with a link is required.","still_real":null}