Sep 23, 2026 · 46 min
AI companies knew search could drain news traffic
OpenAI Admits AI is Killing the Internet
The episode connects private acknowledgments about copyright and publisher losses to a separate weakness that lets synthetic music reach real artists’ audiences.
- 1Lawsuit filings suggest Microsoft and OpenAI understood AI search could reduce traffic to news publishers.
- 2Internal descriptions of training as unprecedented theft clash with the companies’ public defenses of their practices.
- 3A distributor’s weak artist verification let an AI-generated song appear on a real band’s Spotify profile.
Don't miss
Emanuel Maiberg explains how an AI-generated punk song was accepted on Spotify under a real band’s name after minimal distributor verification.
The brief
Newly public filings in The New York Times’ lawsuit portray Microsoft and OpenAI as aware that AI search summaries could divert traffic from news publishers while relying on copyrighted work for training.
The hosts examine internal descriptions of language-model training as an unprecedented theft of labor, then discuss evidence that OpenAI employees developed a way around The New York Times’ paywall.
Emanuel Maiberg tests a separate platform failure by generating a punk song with Udio, creating artwork with ChatGPT, and uploading the track under a real Brooklyn metal band’s name.
The song reaches Spotify through DistroKid with almost no verification, exposing a gap between distributors, streaming platforms, and artists’ need to protect their catalogs.
The episode’s through line is accountability: companies can recognize the harms created by AI systems internally, yet responsibility for preventing those harms remains diffuse.
Featuring
The New York Times
Microsoft Corporation
DistroKidListen to the full episode and explore every guest, topic, and moment on PodLume.