Two of America's Biggest Companies Are Fighting Over Who Actually Trains AI

Keypoints:

• Unsealed court documents show internal Microsoft debate over AI training and copyright
• A Microsoft executive called the practice a possible "largest theft of labor in human history"
• Microsoft says the comment reflected an intentionally provocative internal perspective, not company policy
• Anthropic co-founder Jack Clark, then at OpenAI, separately warned AI could substitute for cultural labor
• The documents surfaced in an ongoing copyright lawsuit against OpenAI and Microsoft

Internal Microsoft documents unsealed this week in an ongoing copyright lawsuit reveal a sharper internal debate over AI training than either company has acknowledged publicly. Brent Hecht, Microsoft's Director of Applied Science, is quoted calling the mechanism behind training AI models on news articles possibly "the largest theft of labor in human history" and "a complete mockery of the idea of 'fair use.'"

Microsoft's official position in the case remains that training AI on news content qualifies as fair use, since the resulting output is different enough from the original articles to count as new work rather than a substitute for it. 

The company said Hecht's comment reflected his role presenting "divergent and asymmetric perspectives" internally, not the company's actual legal or policy stance. Plaintiffs in the case counter that OpenAI used tools specifically built to get past news site paywalls while gathering training data, a detail they argue undercuts the fair-use defense regardless of internal disagreement.

A second document adds another internal voice to the debate. In a memo to OpenAI's Sam Altman and Greg Brockman, Jack Clark, then OpenAI's policy director and later a co-founder of Anthropic, warned the technology was "going to increasingly lead to us creating systems that substitute for the labor of the people that define the 'culture' of society." 

Taken together, the unsealed material shows people inside both companies raising the exact concerns their employers are now fighting in court to dismiss, a gap between private debate and public legal defense that plaintiffs are likely to lean on heavily as the case moves forward.