The Biden administration has intensified its commitment to AI safety and ethical standards. At the UK AI Safety Summit, Vice President Kamala Harris outlined new initiatives to mitigate AI risks, emphasizing the dual nature of AI as a tool for both unprecedented good and potential harm.
Central to these initiatives is the establishment of the United States AI Safety Institute (US AISI) within the National Institute of Standards and Technology (NIST). The US AISI will develop guidelines and best practices for evaluating AI systems, focusing on preventing dangers like AI-driven discrimination and ensuring transparency.
The administration’s proactive stance extends to the legislative domain, with the Office of Management and Budget (OMB) releasing draft policy guidance on government AI use. This document, open for public commentary at ai.gov/input, is a precursor to comprehensive safeguards across various public sector applications.
Moreover, the administration is addressing the misuse of AI in military applications. The Political Declaration on the Responsible Use of Artificial Intelligence and Autonomy, signed by 30 nations, establishes norms for military AI deployment. This is a significant step toward global consensus, though many more nations need to join.
Content authentication is also a priority, with the Commerce Department leading efforts to validate government-produced content in collaboration with industry groups. This includes establishing industry norms and advocating for global standards in content authentication.
Vice President Harris called for international cooperation and legislative action to bolster AI safety, highlighting the need for regulation to ensure technology companies prioritize public well-being and democratic stability over profit.
Taken together with the President’s AI-focused EO earlier this week, I think it’s fair to say that AI is on the fast track to an insane amount of “regulatory capture.” If you are unfamiliar with the term, take a few minutes to watch my friend Bill Gurley’s excellent talk at the All-In Summit.
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various AI models, including but not limited to: GPT-4, Bard, Claude, Midjourney, Stable Diffusion, and others.