Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government
The order establishes nine binding principles — covering accuracy, transparency, safety, civil rights, and accountability — that all covered federal agencies must follow when designing, developing, acquiring, or using artificial intelligence, excluding national security and defense applications.
It creates a cascading implementation process: agencies must inventory their AI systems, assess them against the new principles, and develop compliance or retirement plans for any that fall short, with OMB issuing common policy guidance to govern the full effort.
What this order does
What it orders
The order directs all covered federal agencies — excluding the Department of Defense and the Intelligence Community — to adhere to nine principles when designing, developing, acquiring, or using artificial intelligence: that AI must be lawful, purposeful, accurate, safe and secure, understandable, responsible and traceable, regularly monitored, transparent, and accountable. It requires OMB to publish a roadmap for AI policy guidance, requires each agency to designate a responsible official, build a public inventory of its non-classified AI use cases, and develop plans to bring non-compliant systems into conformance or retire them. It also directs GSA's Presidential Innovation Fellows program to create an AI expertise track and directs OPM to survey federal rotational programs that can build AI workforce capacity.
The order explicitly excludes defense and national security AI, AI embedded in common commercial products such as word processors, and AI research and development activities. It does not itself impose any requirements on the public, creates no enforceable individual rights, and is subject to available appropriations.
Who it affects
All federal executive agencies except the Department of Defense and Intelligence Community components, their employees responsible for AI design and procurement, AI vendors and contractors building systems for federal missions, and members of the public who interact with AI-powered government services such as fraud detection, grant processing, or regulatory review.
Why it matters
Agencies already deploying AI — for tasks like fraud detection, grant reviews, and weather modeling — must now inventory and assess those systems against binding principles. Any system found non-compliant must be fixed or shut down within roughly a year, increasing accountability for how government AI affects ordinary people's benefits, rights, and services.
What must happen and when
How the order is supposed to work
Implementation unfolds in waves. Within 30 days, each agency designates a responsible AI official. Within 45–60 days, the CIO Council publishes interagency coordination forums and the criteria for AI inventories. Agencies then have 180 days from that point to complete their inventories, followed by 120 days to develop compliance or retirement plans and 180 days to implement them. OMB runs a parallel track, publishing a policy-guidance roadmap within 180 days of signing and updating Circulars over time. OPM and GSA's Presidential Innovation Fellows program build AI workforce capacity on separate tracks. No enforcement mechanism beyond agency approval of plans is specified; the order notes implementation is subject to existing resource levels.
Actions and deadlines
- Each agency designates responsible official(s) to coordinate AI principles implementation
- CIO Council publishes list of recommended interagency bodies and forums for AI coordination
- OPM, in coordination with GSA, creates inventory of federal rotational programs for expanding AI expertise
- CIO Council identifies and publishes criteria, format, and mechanisms for agency AI use case inventories
- Presidential Innovation Fellows program establishes an AI expert track to place industry and academic experts in agencies
- OMB Director publicly posts a roadmap for AI policy guidance OMB intends to create or revise
- OPM issues report with recommendations for using rotational programs to expand agency AI expertise
- Each agency prepares an inventory of non-classified and non-sensitive AI use cases, annually thereafter
- Agencies share their AI use case inventories with other agencies through CIO and Chief Data Officer Councils
- Agencies develop compliance or retirement plans for AI applications found inconsistent with the order
- Agencies make their AI use case inventories publicly available
- Agencies implement approved compliance or retirement plans for non-conforming AI applications