US‑China AI Open‑Weight Debate Splits Big Tech Over Security and Openness

American AI firms are divided over the prospect of open‑weight Chinese models. OpenAI and Anthropic argue that the most powerful systems pose safety risks and should remain under tight control, warning that unrestricted access could threaten national security. They cite recent lawsuits over massive copyright infringements in model training and call for regulatory action to curb the rise of Chinese AI competitors.

At the same time, a coalition that includes Microsoft, Nvidia, Meta, IBM and Palantir backs an open‑source approach, saying it fuels innovation, competition and independent security reviews. The group signed a July 24 letter urging US policymakers, including Treasury Secretary Scott Bessent and President‑level adviser Michael Kratsios, to preserve the ability to download and use advanced models without restriction. The dispute reflects broader US‑China tensions as Chinese firms accelerate AI development, prompting both industry and government debate over the balance between openness and control.