LAS VEGAS — The Biden administration is expediting work to develop an executive order to address risks posed by artificial intelligence and provide guidelines to federal agencies on how it might be used, Arati Prabhakar, director of the White House Office of Science Technology and Policy, told CyberScoop on the sidelines of the DEF CON security conference.
As generative AI tools such as ChatGPT have become widely available, Prabhakar said that President Biden has grown increasingly concerned about the technology and that the administration is working rapidly to craft an executive order that will provide guidance to federal agencies on how best to use AI.
“It’s not just the normal process accelerated — it’s just a completely different process,” Prabhakar said, adding that she’s been encouraged by the urgency federal agencies are treating AI regulation. “They know it’s serious, they know what the potential is, and so their departments and agencies are really stepping up.”
Prabhakar spoke to reporters after visiting the AI village at DEF CON, where thousands of hackers here are participating in a red-teaming exercise aimed at discovering vulnerabilities in leading AI models. Over the course of the conference, attendees have stood in long lines for a chance to spend 50 minutes at a laptop attempting to prompt the laptops into generating problematic content.
Prabhakar’s comments come amid a flurry of work on Capitol Hill and the White House to craft stronger AI guardrails.
Senate Majority Leader Chuck Schumer, D.-N.Y., has begun convening a series of listening sessions aimed to educating lawmakers about the technology and laying the groundwork for a major legislative push to regulate AI.
The White House recently announced a set of voluntary safety commitments from leading AI companies, and a forthcoming executive order is expected to provide additional guidance on how to deploy the technology safely. This week, the White House and the Defense Advanced Projects Research Agency announced that they would launch a challenge aimed at using AI to defend computer systems and discover vulnerabilities in open source software.
Prabhakar said policymakers have a unique opportunity today to harness the benefits and govern the risks of what could be a transformational technology.
“A lot of the dreams that we all had about information technology” have today “come true,” Prabhakar said. “But some nightmares have come with that,” and Prabhakar said a growing realization about the harms posed by technology is fueling a sense of urgency in the federal government to put up guardrails.