2026-09-19 · Fuente：HotHardware
Internal documents from Microsoft and OpenAI revealed in The New York Times' lawsuit include one describing AI web scraping as "an astonishing theft," echoing critics' labor-exploitation charges.
The copyright battle over generative AI has produced an awkward disclosure: internal documents from Microsoft and OpenAI, brought to light through The New York Times v. Microsoft and OpenAI, show that misgivings about web scraping were not confined to the technology's outside critics.
The loudest opponents of generative AI have long characterized the technology as overt labor theft, pointing to how the underlying web scraping and model training actually work. According to materials surfaced in the case, those reservations were shared inside the companies themselves. One internal Microsoft document described the practice as "an astonishing theft." The source report further notes that a Microsoft executive warned that AI web scraping makes a mockery of fair use — the same legal doctrine that AI developers have generally advanced to justify ingesting web-published material.
The significance is procedural as much as rhetorical. Because these documents entered the record via The New York Times v. Microsoft and OpenAI, the companies' own framing is now available to the parties pressing claims against them. For a defense that leans heavily on fair use, internal language describing scraping as theft — and characterizing it as making a mockery of the doctrine — hands plaintiffs persuasive exhibit material and complicates the public narrative that training on scraped web content is a routine, legally settled practice.
For the broader AI compute and data ecosystem, the episode underscores that training-data provenance remains an unresolved legal and reputational exposure. The chasm between the industry's public fair-use posture and the skepticism recorded in internal Microsoft documents gives critics of generative AI a powerful talking point, and keeps the question of who pays for web-scraped training material squarely in front of the courts.
Etiquetas: Microsoft · OpenAI · The New York Times · Fair Use · AI Training Data · Copyright Law
萬安算交所 AIXX · AIXX · Exchange verificado · Ir al mercado