How detection works, how to talk about it fairly, and what we’re learning as language models change. Written by our research, education and product teams.
10 posts
A single accuracy number hides a lot. What it measures, what it leaves out, and the three questions to ask any detector before you trust its score.
Maya Okafor
Head of research · Oct 14, 2026
The best time to set expectations about AI writing tools is the first week of term, not the day a paper gets flagged. A practical plan for teachers.
Daniel Reyes
Oct 9, 2026
Search engines say they care about helpful content, not how it was made. So why do so many AI-heavy sites lose traffic? A look at what moves rankings.
Priya Natarajan
Oct 2, 2026
When a detector flags human writing as AI, someone pays for it. How false positives happen, which writing is most at risk, and how we keep them rare.
Tomás Lindqvist
Sep 25, 2026
Our newest detection model is live for every account. It's better on short passages, lightly edited AI drafts and writing from non-native English speakers.
Jonas Weber
Sep 18, 2026
Twenty thousand documents, nine model families and a strict rule about what counts as human. Exactly how we measure our detector, and where it falls short.
Maya Okafor
Sep 11, 2026
Light edits rarely hide AI writing. Heavy rewrites change the picture. What happens to a score as a person reworks an AI draft, sentence by sentence.
Tomás Lindqvist
Sep 3, 2026
You can't detect your way out of every problem. Six assignment formats, tested by teachers we work with, that make the writing process the point.
Daniel Reyes
Aug 27, 2026
Vague policies get ignored. A clear one fits on a page, names the tools and says who checks the work. A starting point for newsrooms, agencies and content teams.
Priya Natarajan
Aug 20, 2026
The overall score is only the start. A walkthrough of the prediction map, confidence bands and the patterns that should make you look closer.
Jonas Weber
Aug 12, 2026
Research notes, classroom guides and product updates, written by the team. Nothing else.