Microsoft and OpenAI lose fight to hide internal documents admitting AI scraping is theft

Microsoft and OpenAI have lost a legal battle to keep internal documents confidential, revealing that AI training practices were described as theft by Microsoft executives. The documents, part of a copyright lawsuit filed by The New York Times, show that both companies acknowledged the significant impact of AI scraping on publishers.

The lawsuit, initiated in 2023, centers on the use of AI models like Copilot, which rely on vast amounts of internet data, including content from news articles. The New York Times has argued that this practice violates copyright laws and undermines the value of journalistic work.

According to court filings, Microsoft's position is that the use of AI for training is consistent with copyright law. However, internal communications suggest that executives viewed the scraping of content as a major threat to publications and a form of theft.

The exposure of these internal documents has raised concerns about the legal and ethical implications of AI training practices. Publishers and content creators may face increased challenges in protecting their intellectual property, while companies could encounter heightened scrutiny over their data usage and potential legal liabilities.

As the case moves forward, the outcome could set a precedent for how AI companies handle training data and interact with content creators. The legal battle remains ongoing, with a ruling expected in 2026.