Microsoft exec calls AI scraping the largest theft of labor in history

Microsoft and OpenAI executives exchanged internal emails this week that warned of an “AI doom loop” that could undermine the viability of news organizations. The correspondence, obtained by a media watchdog, shows senior leaders at both companies debating how large‑scale language models, trained on publicly available news content, might eventually erode the very sources they rely on for data. The messages cite a scenario in which AI‑generated articles displace human reporting, leading to reduced newsroom staffing, fewer original stories, and a shrinking pool of high‑quality text for future model training.

The emails also detail proposed safeguards, including potential licensing agreements with news publishers and the development of attribution mechanisms to ensure that AI outputs credit original sources. Microsoft’s chief legal officer emphasized the strategic importance of preserving a healthy news ecosystem for both corporate reputation and the long‑term performance of its AI products. OpenAI’s research director echoed the concern, noting that a decline in original journalism could degrade model accuracy and diversity over time.

Analysts say the exchange underscores growing industry awareness of the reciprocal relationship between AI development and the news sector. While both firms continue to explore commercial collaborations, the disclosed dialogue suggests they are actively seeking ways to mitigate the feedback loop that could jeopardize the sustainability of traditional news outlets.