Definition
Project Mango is an alleged OpenAI/Microsoft training dataset assembly referenced in The New York Times copyright lawsuit filings, containing copies of at least 160,903 unique works from news publishers.
Key Allegations (Court Filings)
- Assembled from Bing Index scraping and third-party datasets
- Microsoft allegedly sold Bing-purchased dataset as OpenAI training data without publisher consent
- OpenAI obtained 1.8M NYT articles from third party bound by non-commercial use agreement
- Part of broader ai-copyright dispute over training data provenance
Project Mango details are plaintiff allegations from court filings, not adjudicated facts.