greCAPTCHA: Verifying True Authorship by Testing Authors' Understanding of Their Manuscripts

As AI-assisted and AI-generated submissions proliferate, author bylines no longer reliably signal genuine expertise, leaving peer review bodies, universities, and educational institutions in urgent need of new gatekeeping tools. Researcher Atoosa Kasirzadeh proposed greCAPTCHA in an extended tweet thread: rather than passively inspecting the text itself, it tests how well authors understand their own manuscripts, in order to flag submissions lacking substantive human involvement. A prototype has been implemented, with automated scoring reaching AUC 0.90 in predicting true authors—a concrete technical approach to the problem of AI ghostwriting.

Confirmed

• greCAPTCHA was proposed by Atoosa Kasirzadeh; its core idea is to assess "the process of intellectual production" rather than only the output text, measuring the construct of "verification capability"—the domain knowledge and reasoning needed to critically evaluate and defend one's own contributions.
• The system dynamically generates multi-level questions for each submitted manuscript and automatically produces an assessment report based on the author's answers.
• The evaluation is grounded in an empirical user study and semi-structured interviews with 31 active researchers; automated scoring reached AUC 0.90 in predicting whether participants were the true authors of a manuscript.

Why it matters

• Traditional text detection is passive and easily bypassed; greCAPTCHA shifts the question to "does the author actually understand their own paper," offering a new paradigm for academic publishing and educational assessment.
• Participants in the user study gave positive feedback, feeling the questioning captured some degree of true authorship, while also pointing out weaknesses and suggesting improvements before deployment—indicating the method still needs refinement but its direction has received initial validation.

2026-09-17 ~ 2026-09-17 · 7 related posts

• [source] AI-Generated Submissions Are Flooding Peer Review — A Researcher Proposes Evaluating Process, Not Prose — Dr_Atoosa · 2026-09-17
• greCAPTCHA: testing manuscript-specific understanding to catch AI-written papers — Dr_Atoosa · 2026-09-17
• greCAPTCHA Proposes Measuring Whether Authors Actually Understand Their Own Manuscripts — Dr_Atoosa · 2026-09-17
• [source] Implemented greCAPTCHA Dynamically Generates Multi-Level Questions to Verify Manuscript Authorship — Dr_Atoosa · 2026-09-17
• [source] greCAPTCHA verifies manuscript authorship with 0.90 AUC in user study — Dr_Atoosa · 2026-09-17
• Automated authorship scoring hits AUC 0.90 in verifying real manuscript authors — Dr_Atoosa · 2026-09-17

1 near-duplicate retellings: Dr_Atoosa