AI Text Classifier
ExternoThe OpenAI AI Text Classifier was a free web tool launched in January 2023 to detect AI-generated text, providing labels from 'very unlikely' to 'likely AI-generated' for inputs over 1,000 characters. Trained on diverse human English text and outputs from 34 AI models across five organizations, it offered a simple way for users to assess text authenticity amid rising AI content creation. Though innovative, it was discontinued in July 2023 due to low accuracy, marking an early but imperfect step in AI detection technology.
Descripción
The OpenAI AI Text Classifier was a free web tool launched in January 2023 to detect AI-generated text, providing labels from 'very unlikely' to 'likely AI-generated' for inputs over 1,000 characters. Trained on diverse human English text and outputs from 34 AI models across five organizations, it offered a simple way for users to assess text authenticity amid rising AI content creation. Though innovative, it was discontinued in July 2023 due to low accuracy, marking an early but imperfect step in AI detection technology.
Capacidades clave
- Detect AI-generated English text with likelihood labels (very unlikely, unlikely, unclear, possibly, likely AI-generated)
- Classify long-form prose (minimum 1,000 characters) as human or AI-written
Casos de uso principales
- 1.Basic flagging of AI-generated content in documents
- 2.Tentative authenticity checks for editors and researchers
¿Es AI Text Classifier para ti?
Ideal para
- Editors and researchers needing basic, tentative AI checks on long English prose
No ideal para
- High-stakes users like educators relying on accurate detection
- Non-English languages, short texts, or code
- Plagiarism verification or advanced analysis needs
Funciones destacadas
- Free open-access web interface
- No login required
- Simple paste-and-submit with quick results
- Trained on human text and 34 AI models from 5 organizations
Precios
Quarterly Starter
Smart Essential
Super Saver
Feedback Destacado
Puntos Fuertes
- Free and easily accessible
- Intuitive, user-friendly interface
- Backed by OpenAI's diverse training data
Quejas Comunes
- Low accuracy: 26% true positives, 9% false positives
- Often returns 'unclear' results
- Unreliable for short texts, non-English, code, or edited AI text
- Discontinued July 20, 2023
- Lacks plagiarism detection or advanced features