The Government Accountability Office (GAO) published a report this week saying that Federal agencies are hitting the requirements set by President Biden’s October 2023 AI executive order (EO) so far. However, GAO officials are letting agencies know that they expected this outcome and will be watching to make sure agencies continue to get it right.
The report looked at 13 requirements of the AI EO for what agencies should have implemented by March 2024 – the sixth-month milestone. Agencies fully met these 13 requirements, GAO said, laying the groundwork for government-wide AI efforts.
“It was a good news report. Rarely for GAO, we had no recommendations. We found that the agencies did everything all right, which makes sense,” Kevin Walsh, a director of IT and cybersecurity at GAO, said on Thursday during the Government & AI Summit hosted by Nextgov/FCW.
“You’d think that the low-hanging fruit would be the things that the executive office would put in earlier, rather than later,” Walsh added. “But the bottom line of that is, ‘Hey, we’re watching, we’re making sure that agencies are doing the right thing, and we’re looking at AI.’”
Walsh said the report also provides updates on AI-related hiring, noting that the White House has already set up the interagency Chief AI Officer Council, “so that we can coordinate, not reinvent 26 different wheels, but try to come up with some good ideas.”
While the report offers a positive outlook on Federal agencies’ AI progress thus far, Walsh said that the government still has work to do when it comes to preparing to implement AI.
“AI is here. It’s been around for a while, and the government is already using it,” Walsh said. “There is a lot of potential that we see, but the government needs to get on board and really get things set up, get that infrastructure … set up.”
Similarly, Dave Hinchman, who is also a director of IT and cybersecurity at GAO, said last week that in conducting the new report, GAO learned “that the government is just now beginning its AI journey as a whole.”
“I don’t think there’s as much AI really out and deployed in the Federal government as we think there is,” Hinchman said on Sept. 4 at the Billington CyberSecurity Summit in Washington. “One hundred and fifty days is not a lot of time for a huge, sprawling, federated agency to really get in and do it, but this is all part of the journey.”
“There’s a lot that the government is still struggling to understand. There aren’t a lot of policies in place. There aren’t a lot of frameworks and structure around that,” he added. “And so, I think that over the next several years, as AI becomes more commonplace, as agencies start to figure out what they’re going to do with it, I think we’re going to really start to see some interesting things – both in terms of the threats that agencies face, as well as the infrastructure sectors, but also the powerful tools that AI can be used for in terms of fighting those threats.”
