Microsoft—Unsealed court documents revealed Microsoft ran secret OpenAI web-scraping project despite internal 'doom loop' warnings
Documents unsealed in October 2026 in the consolidated OpenAI copyright MDL (S.D.N.Y., No. 1:25-md-03143) showed Microsoft directed employees to keep confidential a project supplying OpenAI with billions of scraped webpages (internally called 'Project Taxi') and a crawler built for OpenAI ('Project Mango'). Internal Microsoft and OpenAI documents showed both companies recognized their products were causing steep drops in referral traffic to publishers (as much as 83-93% for some sites) and risked creating a 'doom loop' that could gut the web's economics, while continuing the practices and keeping the scraping arrangement secret.
Scoring Impact
| Topic | Direction | Relevance | Contribution |
|---|---|---|---|
| Corporate Transparency | -against | primary | -1.00 |
| Intellectual Property Ethics | -against | secondary | -0.50 |
| Overall incident score = | -0.664 | ||
Score = avg(topic contributions) × significance (high ×1.5) × confidence (0.59)
Evidence (1 signal)
Bloomberg Law: Microsoft told staff to keep OpenAI scraping project secret
Newly unsealed filings in the OpenAI copyright MDL showed Microsoft instructed employees to keep confidential its 'Project Taxi' and 'Project Mango' arrangements supplying OpenAI with scraped web content, despite internal recognition of a 'doom loop' harming publishers.