Executive Order 14409: Promoting Advanced Artificial Intelligence Innovation and Security
President: Donald J. Trump
Directed agencies to support frontier AI research and infrastructure while improving evaluation, cybersecurity, supply-chain and national-security safeguards.
Status: In effect.
Is Executive Order 14409 still in effect? Yes. Active; advanced-model security and innovation program directed.
What does Executive Order 14409 do?
The text directs NIST and Commerce to develop evaluations, standards and measurement tools, and Energy and Defense to support secure research and computing.
Reading the operative provisions of Executive Order 14409
EO 14409 directs agencies to support advanced AI research and infrastructure while strengthening evaluation, cybersecurity, supply-chain and national-security safeguards. It assigns NIST, Energy, Defense and other departments roles in coordinating research, computing and standards. The Federal Register text establishes federal priorities; it does not approve a specific model or exempt AI developers from existing law.
Frontier AI systems require substantial computing resources and can create risks involving cyber operations, sensitive data, reliability and misuse. Federal agencies can support testing and standards, but evaluations must state what a system was tested to do and what limitations remain. A benchmark score does not establish safe behavior in every real-world setting.
The order’s innovation goals operate alongside privacy, civil-rights, procurement, export-control and security requirements. Agencies must protect classified and sensitive information when sharing compute or data. The implementation record should include funded programs, evaluation protocols, cybersecurity controls and procurement standards. This allows readers to distinguish federal research support from an endorsement or safety certification of any commercial model.
Advanced AI infrastructure can support scientific research and national-security work, but shared computing environments need access controls and monitoring. Agencies should define who can use the systems, which data may be uploaded and how model outputs are evaluated. Sensitive applications may require separate testing from public-facing uses. A coordinated research program should report both capability gains and unresolved risks so that innovation claims remain grounded in evidence.
Before and after
Before the order
Advanced AI development was governed through export controls, procurement, voluntary technical standards, sector law and private security practices.
After the order
Agencies were directed to develop evaluations, secure infrastructure and research partnerships; no general private-model license was created by the order alone.
Federal Register, agency, amendment, revocation and court sources were searched through September 15, 2026. No revocation or material order-specific final court ruling was located; standards, evaluation, infrastructure and security work underway.
Who had to do what
| Agency / office | Action | Timing / position |
|---|---|---|
| NIST and Commerce | Develop evaluations, standards and measurement tools | Ongoing |
| Energy and Defense | Support secure research and computing | Ongoing |
| Homeland Security | Coordinate critical-infrastructure risk work | Staged deadlines |
What happened next
EO 14409 signed.
Legal and source detail
How it works legally
Export-control, procurement, cybersecurity, privacy, civil-rights, intellectual-property and sector statutes apply. Voluntary standards become binding only through a lawful instrument.
Sources by type
If a source is unavailable, first check the cited publisher or another official repository. The cited publisher remains the authority.
Primary and official sources
Orders, laws, court records and official implementation material.
- NIST — AI Risk Management Framework (opens in a new tab) Official record
How we checked this
Last checked against the sources above on September 15, 2026. How we research →