White House Urges Agencies To Adopt AI for Military, Spy Use

(Bloomberg) -- The Biden administration said the US must accelerate adoption of artificial intelligence for military and intelligence uses, directing agencies to quickly deploy the most powerful systems in a safe manner, according to a new national security memorandum.

Announcing the strategy, National Security Advisor Jake Sullivan said the US must extend its lead on AI while also managing the risks, arguing the fast-evolving technology could impact everything from nuclear physics to rocketry and stealth — the ability of aircraft and other vehicles to avoid detection.

“We have to be faster in deploying AI in our national security enterprise than America’s rivals,” Sullivan said in a speech to national security leaders at the National War College in Washington on Thursday. US adversaries are in “a persistent quest to leapfrog our military and intelligence capabilities,” he said, while warning they’re unlikely to be bound by the same principles over its use. 

The ambition to deploy AI in ways that would include lethal force is driven by fears that the US risks falling behind China’s efforts to rapidly advance its own cutting-edge efforts, according to a senior administration official who briefed reporters on the memo before its release on Thursday and requested anonymity to discuss it.

Related:The New Frontiers of Cyber-Warfare: Insights From Black Hat 2024

Sullivan said US advantages in areas such as space and undersea operations could “vanish” with AI-enabled technology, and he urged national security leaders to work “more closely” with AI developers to use powerful systems.

Despite a recent string of meetings with Chinese counterparts where areas of cooperation on AI have emerged, Sullivan said those discussions failed to ease deep US concerns about the ways in which China “continues to use AI to repress its population, spread misinformation, and undermine the security” of the US and its allies.

“Our competitors want to upend US AI leadership and have employed economic and technological espionage in efforts to steal US technology,” according to a White House fact sheet on the memo. It added that the government would provide AI developers with timely cybersecurity and counterintelligence information “necessary to keep their inventions secure” and intervene to improve the security and diversity of chip supply chains.

Responding to the new strategy, Sarah Myers West, co-executive director of the AI Now Institute, a policy research center, warned against rushing to adopt the technology across many different domains, arguing for time-tested approaches to ensure safety. She said there’s a risk that in life-and-death decisions, human military operators may defer to the recommendations of AI systems.

Related:Palantir Wins $100 Million U.S. Contract for AI Targeting Tech

A Defense Department directive published last year called for the US to develop autonomous and semi-autonomous weapons systems that allow commanders and operators to exercise “appropriate levels of human judgment over the use of force,” a more minimal commitment than always having a human decision-maker “in the loop.” 

US declarations over how it views the responsible military use of AI and autonomy have garnered support from more than 50 countries, but they have repeatedly fallen far short of demands from groups such as the Stop Killer Robots campaign, which called it “feeble.”

United Nations Secretary General Antonio Guterres is leading the call for nations to sign a ban on lethal autonomous weapons by 2026, something that the US is not expected to join. 

The administration official said the US national security community was acutely aware of safety concerns relating to the use of AI on battlefields and for intelligence purposes, adding that US AI systems must go through an accreditation process and that there will likely be challenges to adopting the new technology. 

The US is already using AI for help with identifying targets as part of a long-running project to develop algorithmic warfare and is developing hundreds of other AI-related defense projects. Adherents say AI targeting reduces time spent by analysts and can help speed up the number of targets the US military can strike in a day.

Mark Milley, a former chairman of the Joint Chiefs of Staff, described military AI as a “double-edged sword” in a speech this week to help inaugurate the Vanderbilt University Institute of National Security. Previewing his expectation that the future of war will turn on AI and robotics, he also warned the prospect opens up a “Pandora’s box.”

The memo also emphasizes bolstering the work of AI researchers at universities and other organizations that are not drawn from the technology industry. “AI is moving too fast, and is too complex, for us to rely exclusively on a small cohort of large firms,” the White House said in the fact sheet. 

The memo builds on an executive order signed last year by President Joe Biden as well as instructions for the use and procurement of AI technologies by agencies not involved in national security matters. That previous guidance identified specific AI risks to safety and human rights, including loss of life, which contrasts with some of the imperatives of defense and intelligence agencies.