OpenAI, Microsoft Warned Internally of Web 'Doom Loop'

3 articles • Started about 2 months ago • Last updated about 2 months ago

OpenAI and Microsoft acknowledged internally that their large language models are creating a "doom loop" that damages the publishing industry and the web itself, according to reports published September 17-18. Internal discussions at both companies revealed concerns that AI systems trained on web content are degrading information quality online while potentially harming publishers whose work trained the models. The warnings suggest leadership at both firms understood the cyclical damage their technology could inflict on the internet ecosystem that sustains their operations.

Thursday, September 17, 2026

Friday, September 18, 2026