THE WORLD IS NOT STANDING STILL.RSS
BIG CHANGE.
BIG CHANGE, one tap away
BIG CHANGE, one tap away

Make room for the big picture.

Add BIG CHANGE to your home screen or desktop. It opens in its own window, with the latest stories when you’re online.

iPhone & iPad
Open bigchange.ai in Safari. Tap Share, then Add to Home Screen. Turn on Open as Web App if shown, then tap Add.
Android
Open bigchange.ai in Chrome. In the browser menu, choose Install app or Add to Home screen.
Desktop
In Chrome or Edge, use the install icon in the address bar or the browser’s app menu. In Safari on Mac, choose File → Add to Dock.

Markdown edition

# DetectorScope exposes the features behind AI-text classification

> DetectorScope exposes feature contributions in AI text classification. Its research detector does not establish individual authorship.

By BIGCHANGE.AI Editorial

Published: 2026-10-25T10:28:38.470Z
Updated: 2026-10-25T10:28:38.470Z
Canonical: https://bigchange.ai/blog/detectorscope-auditable-ai-text-classification

![Curving paper ribbon with three orange sections over charcoal strokes.](https://bigchange.ai/api/media/file/detectorscope-feature-ribbon-hero-v2.png)
AI-generated conceptual illustration by BIG CHANGE; no actual DetectorScope interface, classifier output, authorship case or BIG CHANGE run is depicted.

DetectorScope's [paper](https://arxiv.org/html/2610.30006v1) was submitted October 21. The [public repository](https://github.com/google-deepmind/detectorscope) documents code, models and a demo; its October 25 news entry announces a forthcoming EMNLP appearance without dating those resources' release.

## The big change

- **What changed:** Named features explain the classifier's score.
- **Why it matters:** Researchers can inspect questionable formatting cues behind classifications.
- **What to watch:** The explanation cannot establish authorship. The repository rules out individual decisions based on these research detectors.

## Method

Sparse autoencoders extract features from a language model's activity. Generated descriptions are checked against feature activations; poor matches are excluded. A sparse classifier reports how each remaining feature contributes to the score. For input text, the documented API returns a label, probability, feature contributions and optional token highlights.

## Evidence

The [study](https://arxiv.org/html/2610.30006v1) tested English benchmark texts and peer reviews, including held out generators. Targeted paraphrases generally changed scores as predicted; repeated guided edits flipped many labels.

Descriptions can miss activation conditions, and the study used one Gemma based model family. The repository cautions that its research detectors err on human text and must not decide individual cases.

BIG CHANGE's [earlier watermark coverage](https://bigchange.ai/blog/openai-textgrain-text-watermark-rollout-limits) examined a different kind of text-origin signal. The repository documents a Python package, downloadable models and a demo. BIG CHANGE has not run them or verified deployment cost. Individual authorship requires evidence beyond a detector score.

## Sources

- [Building Auditable AI Text Detectors with Sparse Autoencoders, arXiv v1](https://arxiv.org/html/2610.30006v1) — Method, tests and limits in the authors' study.
- [DetectorScope official repository](https://github.com/google-deepmind/detectorscope) — Documented code, model, demo and install paths; its disclaimer excludes decisions about individuals. The dated news entry announces a forthcoming conference appearance, not a resource release date.