Top managers of Microsoft acknowledged the illegality of data collection for training AI

During the court proceedings, unedited documents became available in which the management of Microsoft openly acknowledges problems with using data to train neural networks. The company's top managers characterized the practice of mass parsing of content as "the largest theft of labor in human history".

According to the case materials, internal discussions within the corporation indicate an awareness that current data collection methods create a "death loop" for publishers. The documents directly state that AI models threaten the viability of the media business, as systems are trained on paywalled content without obtaining the necessary licenses or paying royalties to copyright holders.

The disclosure of this correspondence undermines the company's legal position in current antitrust and copyright disputes. Previously, corporate representatives insisted that the use of publicly available data on the internet falls under the doctrine of fair use, but internal risk assessments indicate that there are serious doubts about the legitimacy of such actions at the level of top management.

The publication of these data caused a wide resonance in the professional community, including discussions on Hacker News. The situation may lead to a revision of data licensing standards for training large language models and increased pressure from regulators demanding transparency in intellectual property matters.