AI & Tech Brief: Anthropic’s leaked bill

A draft proposal written by Anthropic suggests that AI companies should be legally required to report safety testing results if they spend more than $1 billion on training a single model. The document aims to establish federal oversight standards for frontier AI development by tying regulatory scrutiny to the financial resources invested in compute power. By setting this monetary threshold, the framework seeks to create a predictable standard for identifying high-risk systems before they are deployed to the public.