US urges G20 countries to allow AI training on creators' work

Chapel Hill, North Carolina, on September 2, the United States called upon G20 nations to establish guidelines for how AI corporations utilize copyrighted material to train their systems. This comes amid numerous lawsuits filed by authors, publishers, media organizations, and other creators against major American tech firms like Anthropic, OpenAI, Google, and Meta.

These companies are accused of misusing their work to develop AI products. Tech firms train their AI models by feeding them vast quantities of data, which often include books, music, and art. The firms argue that using this work to build their AI tools is lawful. U.S. Commerce Secretary Howard Lutnick, speaking at a tech-focused meeting in North Carolina, suggested that AI companies should be able to train their models on creators' work while protecting artists.

He advocated for frameworks that safeguard creators and inventors without obstructing the next wave of innovators. The U.S. Justice Department also filed a brief supporting OpenAI in its lawsuit against the New York Times and other newspapers over the latter's use of their work to train OpenAI's AI language models, asserting that AI training generally qualifies as fair use of copyrighted material.

Nvidia CEO Jensen Huang urged G20 officials to focus on real-world problems related to AI rather than theoretical harms when developing regulations. He emphasized the potential negative impact of government requirements on industry profits, as they might slow the release of new models or force companies to alter their products' performance to address security concerns.

Other speakers included Anthropic co-founder Tom Brown, Palantir's Alex Karp, and OpenAI's Sam Altman.