The Office of Management and Budget (OMB) unveiled its consolidated 2024 Federal AI use case inventory this week – noting that agencies have reported over 1,700 use cases – more than doubling the 757 AI tools reported in 2023.
The top three categories for AI use cases in the new inventory are: mission-enabling; health and medical; and government services.
Under the Trump-era executive order on AI, Federal agencies are required to make public their AI use case inventories. In March, OMB released its first-ever governmentwide AI policy, giving agencies a Dec. 1 deadline for inventorying and sharing their AI use cases with the public.
The 2024 inventory, which is available on GitHub, includes 227 use cases that agencies identified as rights-impacting and or safety-impacting.
Two hundred and six of the rights-impacting and safety-impacting AI use cases received an extension for up to one year in order to achieve compliance with OMB’s risk management practices.
According to OMB, the most commonly cited risk management practices that agencies requested extensions for include the requirement to conduct independent evaluations, mitigate emerging risks to rights and safety, and complete an AI impact assessment for their rights- and safety-impacting use cases.
Thirty-seven agencies submitted their Federal AI use cases, with the Departments of Health and Human Services (HHS) and Veterans Affairs (VA) being the only agencies to report more than 200.
The VA also noted that 145 of its 229 use cases are safety and rights impacting – far exceeding the Department of Homeland Security, which was ranked the second highest for 34 safety and rights impacting AI tools.
All reported agency AI use cases, with some wide exceptions, are included in this inventory. Certain use cases are not required to be individually inventoried or publicly released. Excluded categories include research and development use cases; national security systems; intelligence community systems; and Pentagon use cases.
