A New York Times report dated 17 September 2026 says Microsoft executives raised concerns about OpenAI’s alleged harvesting of published material for AI training. The claims appear in court filings from an ongoing copyright-infringement lawsuit involving Microsoft and OpenAI; the article says only partial excerpts are publicly available, so their full context is not known.
According to the excerpts, Brent Hecht, Microsoft’s director of applied science, described the alleged data harvesting as resembling “the largest labour theft ever recorded”. Nick Turley, an OpenAI vice-president, reportedly acknowledged that the practice could pose an existential threat to publishers. Authors and publishers argue that technology companies use large datasets without permission or payment, potentially undermining the funding needed to produce journalism and other content.
The filings allegedly claim that OpenAI continued acquiring material despite awareness of infringement risks, including using technical methods to bypass website paywalls and removing copyright notices from premium content. They further allege that employees were encouraged to pursue such methods, with OpenAI president Greg Brockman reportedly approving a paywall-attack technique in an email exchange.
The article also cites a 2023 observation by an engineer that users rarely click source links in AI-generated responses, and says Turley suggested AI products could largely replace traditional journalism. These remain allegations drawn from incomplete legal documents, rather than confirmed court findings.