OpenAI and Microsoft acknowledged risks to the web ecosystem due to data collection

Declassified court documents of OpenAI and Microsoft have confirmed that the companies' management was aware of the negative impact of their model training strategy on internet infrastructure.

In internal discussions, employees used the term "doom loop" to describe the process by which AI systems consume website content, depriving them of traffic and incentives to create new material. In correspondence, data collection for training neural networks was directly referred to as "the largest labor theft in human history".

These documents have become part of the proceedings related to lawsuits by copyright holders and content publishers. Internal assessments indicate that the companies understood that the aggressive use of data without compensation to authors undermines the economic model of the open web, on which the operation of search engines and AI products themselves is based.

Discussion of these materials is actively underway in the Hacker News community, where experts are analyzing the legal implications of these admissions for future data licensing deals. The disclosure of this information may affect the outcome of current court cases on copyright issues and the use of content to train large language models.