US targets AI risks with mandatory reporting

The news: The US proposed new rules requiring advanced AI developers and cloud providers to report detailed activities to the federal government to ensure AI technologies meet safety and reliability standards.

The Commerce Department’s Bureau of Industry and Security (BIS) on Monday said the proposed rule will cover the development of "frontier" AI models, cybersecurity measures and outcomes from so-called red-teaming efforts designed to test their potential misuse in cyberattacks or aiding the development of weapons of mass destruction.

The context: It follows a 2023 executive order by President Biden to safeguard AI systems critical to US security and follows stalled legislative efforts in Congress.

The BIS said it had conducted a pilot survey earlier this year and is now moving to formalise the reporting requirements.