The Department of Homeland Security faces steep challenges in carrying out the Biden administration’s artificial intelligence playbook, which places the agency at the center of the action.
DHS is charged with developing security guidelines, using the technology to shore up cyber defenses, and working with energy companies, water utilities, and other critical infrastructure to ward off AI-enabled attacks.
President Joe Biden laid out the responsibilities in a sweeping executive order that affects DHS and a host of other agencies. The department will fall short of its AI security goals if it can’t fill gaps in technical expertise and navigate misgivings among civil liberties advocates.
“I’m worried once again that we’re adding missions to DHS without adding the resources that will allow those missions to succeed,” said former counterterrorism official Tom Warrick, now at the Atlantic Council.
Biden mandated an aggressive timeline for the agency’s work: 180 days to weave security considerations into practical guidelines for critical infrastructure, test how AI can help address vulnerabilities in government networks, and study how it can be misused to create weapons of mass destruction, among other tasks. Department leaders are projecting confidence that they can meet the moment.
“We have such a diversity of skill sets and expertise across the department that not only do we have the bandwidth to do this, I don’t know that anyone else has the composition and unique mix of capabilities that we do,” said Eric Hysen, DHS’s chief information officer and its first chief artificial intelligence officer.
Building Expertise
The Cybersecurity and Infrastructure Security Agency, which will do most of DHS’s heavy lifting on the executive order, is already looking to build out its team of AI specialists. CISA released a roadmap Nov. 14 that commits to training the existing workforce on AI systems and recruiting experts to join the 3,000-worker DHS unit.
With the current workforce, DHS and CISA would “fall short on the technical resources needed to provide meaningful impacts,” said Brian Harrell, who was the department’s assistant secretary for infrastructure protection during the Trump administration. Department leaders have yet to specify how many new employees they may need to reach their goals.
“Finding AI talent is hard all around — it’s hard in the private sector, it’s hard in government,” Monument Advocacy lobbyist Andrew Howell said. The talent pool for emerging technology expertise is shallow in the first place, and many workers gravitate to private-sector jobs that generally offer higher pay and a swifter hiring process.
DHS hopes to harness a unique cybersecurity personnel system, established two years ago to streamline cyber hiring, to fill some AI roles, Hysen said. And Biden’s order directs the US Office of Personnel Management to figure out ways to ease hiring for AI jobs across the government.
Biden Signs Sweeping Order Regulating Artificial Intelligence
Hysen credited the AI experts DHS already has and said the agency is “leaning on them very heavily.” DHS uses AI in many ways, including verifying travelers’ identities through facial recognition at airports, helping decide who should go through secondary screening at border crossings, and assessing disaster sites.
DHS is also working closely with tech companies to map out future AI policies. Hysen and Homeland Security Secretary Alejandro Mayorkas met executives from
Collaboration with other agencies — including the Office of Management and Budget and the National Institute of Standards and Technology — will be key to DHS’s success, said Jamie Danker, senior director of cybersecurity and privacy services at Venable LLP and a former DHS privacy officer.
“The pieces that DHS is working on are well suited for them,” Danker said. “There’s a huge number of places the DHS is going to need to partner with other agencies. And so this is going to be a collaborative effort.”
Privacy, Civil Liberties
Among DHS’s resource challenges will be adequate spending and staffing on its privacy and civil liberties programs to oversee a boom of AI use cases.
“The broad set of activities and objectives and the ways that they use AI may put their office in a position where they need to do a lot of work to actually make the case that they are using AI responsibly and in a trustworthy way,” said Heather West, Venable’s senior director of cybersecurity and privacy services and coordinator of the Alliance For Trust in AI, an industry group. “And because of that, it would serve them well to make sure that they are resourced to do this.”
Government watchdogs and advocacy groups have dinged the agency for speeding ahead with new technologies, including AI-enabled facial recognition, without properly vetting privacy and civil liberties harms.
Airport Face Scans Embraced by Agencies Despite Nagging Mistrust
The new demands of the executive order run the risk of accelerating those harms without serious oversight changes, said Faiza Patel, senior director of the Liberty and National Security Program at the Brennan Center for Justice at NYU School of Law.
“When DHS is already using many of these very risky technologies, and that seems to be set to expand, strengthening of oversight structures is even more critical,” Patel said.
Advocates also worry powerful tech companies will have outsize influence over DHS’s AI policies, with resources, personnel, and time to build relationships proactively with the department and get involved in making decisions.
“The onus ends up being on civil society, the least-resourced organizations, to try to weigh in and have their voices heard,” said Corynne McSherry, legal director of the civil liberties group Electronic Frontier Foundation.
Hysen said DHS’s Office of Civil Rights and Civil Liberties and its Office of Privacy are important voices in the agency’s policymaking and technology deployment.
The civil rights office leads the Responsible Use Group within DHS’s AI Task Force. The group spearheaded a directive in September dictating testing of facial recognition for bias and reviews of existing agency uses of the technology. Next it will announce pilots for responsible uses of generative AI, Hysen said.
Looking Ahead
The additional demands on DHS come as congressional turmoil leaves agency budgets in limbo. Lawmakers have twice resorted to stopgap measures with flat funding to keep the government open while they continue to haggle over budgets for the fiscal year that started in October.
“A 2024 budget would be nice,” Hysen said. “We are being extremely flexible with our resources to meet the moment here. AI is such a fast-moving technology that the federal budget cycle, even if it were working as intended, would not be fast enough to keep up.”
The agency is relying in part on a rotating account approved in fiscal 2022 that allows DHS to take a portion of expiring funds every year and use them for information technology and facilities modernization, he said.
DHS leaders haven’t yet said whether they’ll request an infusion of funds for the next fiscal year to meet their AI goals. The administration’s 2025 budget request is expected in February or March.
In the meantime, DHS should focus on building support among lawmakers who’ve been skeptical of CISA, said Howell, the lobbyist. The agency has at times landed in congressional Republicans’ cross hairs, with accusations of free-speech censorship — charges CISA denies.
“In the current political environment, DHS — but particularly CISA — have some work to do in order to rebuild relationships on Capitol Hill,” Howell said. “It’s important that they focus on doing that as part of the additional responsibilities that they’re being given by this AI executive order.”
To contact the reporters on this story:
To contact the editors responsible for this story: