U.S. Proposes AI Reporting Requirements

By David Shepardson

WASHINGTON (Reuters) – The U.S. Commerce Department announced a proposal to enforce detailed reporting requirements for advanced artificial intelligence developers and cloud computing providers. This initiative aims to ensure the safety of technologies against cyberattacks.

The proposal from the Bureau of Industry and Security would mandate reporting to the federal government on the development activities of “frontier” AI models and computing clusters.

Additionally, it requires reporting on cybersecurity measures and outcomes from red-teaming efforts, which test for dangerous capabilities, including those that could assist in cyberattacks or facilitate the development of harmful weapons.

Red-teaming has been a long-standing practice in cybersecurity for identifying new risks. The term originates from Cold War simulations, where the opposing forces were designated as the “red team.”

Generative AI, which can produce text, images, and videos in response to prompts, has generated significant enthusiasm alongside concerns about job obsolescence, election interference, and potential catastrophic impacts.

The Commerce Department states that the proposed information collection is critical to ensuring these technologies comply with rigorous safety and reliability standards, can endure cyber threats, and limit misuse risks by foreign adversaries or non-state actors.

In October 2023, President Joe Biden signed an executive order requiring AI developers whose systems pose risks to U.S. national security, economy, public health, or safety to share safety test results with the government prior to releasing products to the public.

This regulation seeks to establish mandatory reporting for advanced AI models and computing clusters. The move comes as Congressional action on AI has stalled. Earlier this year, BIS conducted a pilot survey of AI developers, while the Biden administration has taken steps to prevent China from harnessing U.S. technology in AI, amid growing security concerns.

Prominent cloud providers include Amazon’s AWS, Alphabet’s Google Cloud, and Microsoft’s Azure unit.