The sweeping order would leverage the U.S. government’s role as a top technology customer by requiring advanced AI models to undergo assessments before they can be used by federal workers, according to three people involved in discussions about the order. The lengthy action would ease barriers to immigration for highly skilled workers, an attempt to boost the United States’ technological edge. Federal government agencies — including the Defense Department, Energy Department and intelligence agencies — would be required to run assessments to determine how they might incorporate AI into their agencies’ work, with a focus on bolstering national cyber defenses.
The White House on Tuesday night sent out invitations for a “Safe, Secure, and Trustworthy Artificial Intelligence” event Monday hosted by the president, according to a document viewed by The Washington Post.
The White House declined to comment. The order has not yet been finalized, so details or plans for timing could change, the people said.
The White House is taking executive action as the European Union and other governments are working to block the riskiest uses of artificial intelligence. Officials in Europe are expected to reach a deal by the end of the year on the E.U. AI Act, a wide-ranging package that aims to protect consumers from potentially dangerous applications of AI. Lawmakers in the U.S. Congress are still in the early stages of developing bipartisan legislation to respond to the technology.
Artificial intelligence presents a key test for the Biden administration, which arrived in Washington following promises to curtail the alleged abuses of Silicon Valley. Nearly three years later, the administration has notched few victories in its efforts to address the harms of social media, privacy abuses and the effects of technology on children. Antitrust enforcers appointed by President Biden have brought high-profile competition lawsuits against some of the most powerful companies in the tech sector, including Amazon, but also suffered from brutal defeats in the courts.
(Amazon founder and former CEO Jeff Bezos owns The Washington Post. The Post’s interim CEO, Patty Stonesifer, sits on Amazon’s board.)
Recent rapid advances in artificial intelligence have raised the stakes, as the launch of ChatGPT and other generative AI tools has accelerated a global movement to regulate U.S. tech giants. Policymakers around the world are increasingly worried that AI could supercharge long-running concerns about tech’s impact on jobs, surveillance and democracy, especially ahead of a critical year for elections around the globe.
The White House first announced plans for the executive order in July, and Biden more recently teased plans for the action in September at a meeting with the President’s Council of Advisors on Science and Technology in San Francisco.
“This fall, I’m going to take executive action, and my administration is going to continue to work with bipartisan legislation so America leads the way toward responsible AI innovation,” Biden said.
Monday’s executive order is expected to build on a set of voluntary commitments signed by 15 companies, including OpenAI, Google, Photoshop maker Adobe and chip maker Nvidia. Brokered by the White House in September, the agreements require the firms to develop technology to identify AI-generated images and include a vow to share data about safety with the government and academics.
The assessments of the government-purchased large language models, exercises known as “red teaming,” are expected to be led by the National Institute of Standards and Technology, a federal laboratory known as NIST that develops standards for new technology, according to two of the people. NIST this year released a framework for managing the risks of AI, and it regularly collaborates with other government agencies on AI.
The government has previously used its purchasing power to push for greater digital controls. A 2021 cybersecurity executive order directed the Commerce Department to craft cybersecurity standards for companies that sell software services to the federal government — a move that officials said they hoped would ripple beyond government contractors and change practices throughout industry and outside the United States.
The immigration provisions follow years of Silicon Valley efforts to push Washington to lower the barriers for engineers and other high-skilled workers, amid fierce competition for technical talent. The Department of Homeland Security signaled this work is already underway. On Monday, the agency announced it was proposing changes to the H-1B program, a high-skilled visa program widely used in Silicon Valley. The State Department will also launch a new program focused on artificial-intelligence talent, one of the people said.
In addition to the changes to the immigration process, federal government agencies would have to take steps to assess the current size of the AI workforce.
Meanwhile, Congress is forging ahead with its own plans for legislation, with the support of the White House. On Tuesday, Senate Majority Leader Charles E. Schumer (D-N.Y.) hosted his second AI Insight Forum, which is expected to serve as the foundation for his bipartisan effort to rein in artificial intelligence. Attendees included venture capitalist Marc Andreessen, AI start-up Cohere co-founder Aidan Gomez and representatives from civil society and labor groups.
The session focused on funding for AI research and innovation, and attendees also expressed the need for changes to immigration processes to attract technical workers. Schumer told reporters that he has sought comprehensive immigration reform for a “long time,” and it remains to be seen whether AI will be different from past efforts.
“There are possible compromises, and some of our members on a bipartisan basis are looking at it,” he said.