July 06, 2023
Frontier AI Regulation: Managing Emerging Risks to Public Safety
Responsible AI innovation can provide extraordinary benefits to society, such as delivering medical and legal services to more people at lower cost, enabling scalable personalized education, and contributing solutions to pressing global challenges like climate change and pandemic prevention. However, guardrails are necessary to prevent the pursuit of innovation from imposing excessive negative externalities on society. There is increasing recognition that government oversight is needed to ensure AI development is carried out responsibly; we hope to contribute to this conversation by exploring regulatory approaches to this end.
We think that it is important to begin taking practical steps to regulate frontier AI today, and that the ideas discussed in this paper are a step in that direction.
In this paper, we focus specifically on the regulation of frontier AI models, which we define as highly capable foundation models that could have dangerous capabilities sufficient to pose severe risks to public safety and global security. Examples of such dangerous capabilities include designing new biochemical weapons, producing highly persuasive personalized disinformation, and evading human control.
This article was originally published by Arxiv by authors Markus Anderljung, Joslyn Barnhart, Anton Korinek, Jade Leung, Cullen O'Keefe, Jess Whittlestone, Shahar Avin, Miles Brundage, Justin Bullock, Duncan Cass-Beggs, Ben Chang, Tantum Collins, Tim Fist, Gillian Hadfield, Alan Hayes, Lewis Ho, Sara Hooker, Eric Horvitz, Noam Kolt, Jonas Schuett, Yonadav Shavit, Divya Siddarth, Robert Trager, Kevin Wolf.
More from CNAS
-
Video
Technology & National Security
New Age of Missiles, Drones Reshapes Battlefields Around the WorldThe world is on the cusp of a third missile age, with conventional ballistic weapons proliferating at unprecedented speed in the wake of widespread, battlefield-warping deploy...
By Paul Scharre
-
Video
Technology & National Security
Spies, Satellites & Startups: Intel in the Digital AgeAs technological innovation accelerates, how should the United States and its allies rethink collaborative threat tracking in an era of commercial tech proliferation? As AI au...
By Anthony Vinci
-
CNAS Insights
Technology & National Security
CNAS Insights | Washington Can’t Afford to Ignore AI’s Warning ShotOn July 21, OpenAI disclosed the first publicly known incident of AI models escaping their isolated testing environment, gaining unauthorized internet access and hacking into ...
By Ruby Scanlon & Janet Egan
-
Commentary
Technology & National Security
Can China Keep Its AI Open?Without reliable ways to measure dangerous capabilities, some judgments may be wrong. And an open model that clears a badly drawn line proliferates globally, offering maliciou...
By Ruby Scanlon
