Microsoft, OpenAI filings admit scraping theft

These unredacted admissions land like a precision strike on the AI industry's fair use defense by supplying direct, executive-level evidence of market substitution, intentional circumvention, and bad faith.

• –**Gutting the Fair Use Defense:** The central pillar of fair use under U.S. copyright law weighs market harm and substitution; internal records showing OpenAI leadership defining ChatGPT as "largely substitutive" alongside Microsoft tracking a 93% drop in referral traffic dismantle claims of purely transformative use.
• –**Evidence of Willful Infringement:** Records showing researchers celebrating paywall bypasses with OpenAI's president ("ah nice") and deliberately removing copyright notices to hide origins severely undermine good-faith protections and expose the firms to statutory damages.
• –**The Publisher Doom Loop:** Microsoft's internal acknowledgment that LLMs are destroying the economic viability of their own "content supply chain" articulates the broader structural crisis facing foundation models: consuming and starving the human knowledge sources they depend on to remain relevant.
• –**Impending Legal Reckoning:** Despite recent amicus support from the federal government favoring fair use for AI training, explicit paper trails demonstrating systemic piracy and paywall evasion make high-value settlements or restrictive judicial rulings far more inevitable.