The Biden administration is releasing the first ever national security memorandum on artificial intelligence, hoping to push more federal agencies into adopting the technology in an effort to maintain any advantage the U.S. has over China and other nations. 

The document “directs the U.S. Government to implement concrete and impactful steps to ensure that the United States leads the world’s development of safe, secure, and trustworthy AI; harness cutting-edge AI technologies to advance the U.S. Government’s national security mission; and advance international consensus and governance around AI.”

Senior administration officials told reporters on Wednesday evening that the memorandum is designed to galvanize federal government adoption of AI and in doing so “shape international norms around AI use to reflect those same democratic values.”

“Failure to do this, a failure to take advantage of this leadership and adopt this technology, we worry, could risk a strategic surprise by our rivals, such as China,” the official said.

Among its provisions, the document calls for protecting the supply of advanced computing hardware, orders the creation of a federal AI Safety Institute, offers guidance on AI governance and risk management in national security missions, and calls for an assessment of how well the U.S. private sector is doing in AI development.

“There are very clear national security applications of artificial intelligence, including areas like cybersecurity and counter intelligence, not to mention the broad array of logistics and other activities that support military operations,” the official said. “Because countries like China recognize similar opportunities to modernize, revolutionize their own military and intelligence capabilities using artificial intelligence, it's particularly imperative that we accelerate our national security community's adoption and use of cutting edge AI capabilities to maintain our competitive edge.”

The officials were pressed by reporters on an array of serious issues around AI — including its use in lethal military targeting operations; companies training systems on troves of American citizen data collected by data brokers; continued hallucinations and false-positive results; and the lack of regulation beyond voluntary agreements.

But members of the Biden administration largely sidestepped these questions, arguing that the increased use of AI among federal agencies would allow them to work through these issues in a way that could provide future guidelines on best practices. 

A White House spokesperson said the administration is trying to set things up so that policy problems can be addressed as quickly as possible.  

“We have a lawyers group that is working very intensively to ensure that all existing law is complied with, but also to ensure that novel legal issues as we encounter them are addressed in a timely way as well,” they said. 

“We want to see rapid adoption, but we also want to see experimentation that will tease out what missions are best suited for various systems, and also play out the challenges with them, and that's going to require leaning forward and experimenting, adopting and then doing all of the work that was just mentioned in terms of both policy and legal review.”

The adoption of AI by federal agencies would be done in a way to ensure that it “reflects democratic values and protects human rights, civil rights, civil liberties and privacy,” they explained, adding that they could “shape international norms around AI use to reflect those same democratic values, and directs actions to track and counter adversary development and use of AI for national security purposes.” 

Administration officials said they have held policy discussions about the memorandum over the last year and said national security adviser Jake Sullivan will discuss the issue further during a speech at a military college on Thursday. 

Aware of the competition

The memorandum includes several provisions around the need for improvements to the “security and diversity of chip supply chains” as well as government supercomputers and other emerging technology. 

Officials said competitors to the U.S. — chiefly China — want to upend U.S. leadership in AI and have “employed economic and technological espionage in efforts to steal U.S. technology.” 

“This NSM makes collection on our competitors’ operations against our AI sector a top-tier intelligence priority, and directs relevant U.S. Government entities to provide AI developers with the timely cybersecurity and counterintelligence information necessary to keep their inventions secure,” officials said. 

OpenAI, one of the leading companies working on large language models, released a report two weeks ago that outlined a cyberattack by nation-state groups from China seeking to steal information from the company. The report also showed that several countries are already using OpenAI tools for disinformation campaigns and malware refinement. 

The memorandum formally designates the AI Safety Institute within the Department of Commerce as U.S. industry’s primary port of contact in the U.S. government. The institute will be staffed by AI technical experts and will “partner with national security agencies, including the intelligence community, the Department of Defense, and the Department of Energy.”  

The document notes that AI is evolving too quickly and has become too complex for it to continue to be controlled by a small coterie of companies and firms — necessitating a more democratized cohort of people and institutions. 

The memorandum “directs the National Economic Council to coordinate an economic assessment of the relative competitive advantage of the United States private sector AI ecosystem.”

The White House said the memorandum will include guidance for AI governance and risk management for use in national security missions. The Office of Management and Budget previously issued guidance for non-national security missions.

In an accompanying document, the Biden administration touted its work this year on gaining some form of international consensus around AI, including several G7 documents, a UN resolution and another declaration on the military use of AI that was signed by dozens nations. 

Free speech and nuclear weapons

The White House plans to publish a companion document that will specifically govern how federal agencies can and cannot use AI. 

When asked for specifics, an administration official noted that there are “clear prohibitions on the use of AI with intent or purpose, for instance, to unlawfully suppress or burden the right to free speech or the right to legal counsel.” 

“There's also prohibited use cases around removing a human in the loop for actions critical to informing and executing decisions by the president to initiate or terminate nuclear weapons deployment, for example,” they said. 

“We actually view these restrictions … as being important in clarifying what the agencies can and cannot do that will actually accelerate experimentation and adoption. One of the paradoxical outcomes we've seen is with a lack of policy clarity and a lack of legal clarity about what can and cannot be done, we are likely to see less experimentation and less adoption than with a clear path for use, which is what the [memorandum] provides.”