AI-Generated Text Detectors Reportedly Misclassified Human and AI-Written Text

January 31, 2023

AI-generated text detectors, including OpenAI’s AI Text Classifier, GPTZero, and Originality.ai, were publicly tested in early 2023 and reportedly produced false positives, false negatives, and easy evasion. According to recent findings, these models struggled to accurately distinguish between human-written and AI-written texts. Notably, OpenAI’s classifier identified only 26% of AI-written samples as likely AI-written and falsely labeled 9% of human-written text. Experts are now warning about the risks of relying on such flawed systems for content moderation, highlighting the need for more robust and responsible AI governance. And to stay informed about AI incidents, visit JOIN US.

Matched TAIM controls

Suggested mapping from embedding similarity (not a formal assessment). Browse all TAIM controls

Alleged deployer
Originality.ai, OpenAI, Edward Tian
Alleged developer
OpenAI, Edward Tian
Alleged harmed parties
Teachers, Students, Educators, Educational communities

AI governance case studies

For forensic AI governance failure analysis (TAIMScore™ case studies), browse Human Signal’s Failure Files™.

Source

Data from the AI Incident Database (AIID). Cite this incident: https://incidentdatabase.ai/cite/466

Data source

Incident data is from the AI Incident Database (AIID).

When citing the database as a whole, please use:

McGregor, S. (2021) Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database. In Proceedings of the Thirty-Third Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-21). Virtual Conference.

Pre-print on arXiv · Database snapshots & citation guide

We use weekly snapshots of the AIID for stable reference. For the official suggested citation of a specific incident, use the “Cite this incident” link on each incident page.