Tracking the U.S. AI Executive Order
On October 30, 2023, President Biden signed Executive Order 14110 on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence (“AI EO”). To follow and assess the federal government’s related actions on AI, in November 2023 we published the Safe, Secure, and Trustworthy AI EO Tracker (“AI EO Tracker”)—a detailed, line-level tracker of the 150 requirements that agencies and other federal entities must implement. Since then, our team of scholars from the Stanford Institute for Human-Centered AI (HAI) and Stanford’s Regulation, Evaluation, and Governance Lab (RegLab) have been continually updating the tracker in line with the government’s efforts to swiftly implement the AI EO’s requirements. Here, we provide an overview of our ongoing work to analyze AI EO implementation.
Access the AI EO Tracker as a public Google Sheet
Download the latest version of the AI EO Tracker as a Microsoft Excel document
Our Latest Tracker Update
Update at 180 Days
The federal government has implemented 36 requirements and made progress on 17 additional requirements since our previous tracker update in early February 2024. In total, the federal government has now implemented 54 of its 150 EO tasks, with another 25 requirements in progress based on our methodology for assessing implementation (see below).
The White House proactively reports its implementation progress, but transparency gaps remain: While its latest fact sheet claims that all requirements due within 180 days have been completed, we could only confirm implementation of 49 (71 percent) of those 69 requirements. We verified another 11 requirements (16 percent) as “in progress,” but couldn’t find conclusive evidence for the remaining nine requirements (13 percent) to confirm claims of completion.
Independently verifying the implementation of EO requirements remains tricky—only a few agencies provide clear, timely, and easy-to-find updates on their EO-related activities. In its fact sheets, the White House also lists some activities as completed that don’t directly correspond with EO requirements.
Background and Methodology
The AI EO Tracker builds upon prior research by Stanford HAI and the RegLab to assess the federal government’s implementation of three previous AI-related executive orders and legislation. Our initial December 2022 implementation tracker and methodology (and accompanying paper) shed light on the accomplishments and challenges in realizing these AI policies. With increased transparency and public accountability through the tracker, we subsequently observed more agencies fulfilling requirements.
Assessing implementation is similarly important for this EO. While the federal government is publicly reporting on its implementation efforts with a level of transparency not seen for prior AI-related EOs, thanks to regular White House fact sheets and frequent agency announcements, we seek to independently verify claims of implementation. Our AI EO tracker provides information on the implementation status of requirements based on 1) the White House’s official fact sheets; 2) announcements, public statements, or official documents issued by the responsible federal entities or officials; and 3) media reports and other conclusive evidence. We mark requirements as:
- “Implemented”: if there is sufficient public evidence of full implementation (e.g., official announcements, reports) separate from White House fact sheets.
- “In progress”: if a requirement has been completed only partially, the extent of progress is ambiguous, or full completion of the requirement necessitates ongoing action.
- “Not verifiably implemented”: if we could not find conclusive, publicly available evidence of full implementation separate from White House fact sheets.
We recognize that there are a number of rationales that counsel against complete disclosure of implementation details, including concerns related to national security and time constraints. However, when public reporting of implementation details do not implicate security concerns, greater transparency is crucial to enhancing public accountability.
Previous Tracker Updates and Analyses
Transparency of AI EO Implementation: An Assessment 90 Days In
The U.S. government has made swift progress and broadened transparency, but that momentum needs to be maintained for the...
By the Numbers: Tracking The AI Executive Order
New Stanford tracker analyzes the 150 requirements of the White House Executive Order on AI and offers new insights into...
Decoding the White House AI Executive Order’s Achievements
America is ready again to lead on AI—and it won’t just be American companies shaping the AI landscape if the White House...