Not sure where to start? Answer 3 questions and get a tailored compliance action plan.
What applies to me? →Promoting Advanced Artificial Intelligence Innovation and Security
Issued by
White House (Executive Office of the President)
This presidential action directs federal agencies to assess the advanced cyber capabilities of frontier AI models and establishes a classified benchmarking process led by the National Security Agency to designate 'covered frontier models.' It creates a voluntary framework through which developers may share frontier models with trusted government partners for security review. Mandatory government licensing or permitting of AI models is explicitly prohibited under the order.
Applies To
Overview
Issued on June 28, 2026, this White House presidential action addresses national security risks posed by the most capable AI systems, referred to as 'covered frontier models,' which are designated through a classified NSA-led benchmarking process based on advanced cyber capability thresholds. The order tasks relevant federal agencies with evaluating whether frontier models meet designation criteria and coordinating security assessments accordingly. A voluntary information-sharing framework is established to allow developers to submit frontier models for government security review without triggering mandatory licensing, permitting, or approval requirements. The order reflects an explicit policy choice to preserve commercial AI development while creating structured government visibility into the most capable systems. Compliance obligations are concentrated among developers of the most advanced models, particularly those with significant cybersecurity-relevant capabilities. Enforcement mechanisms are primarily administrative and interagency rather than punitive, though classified designation processes may carry indirect operational consequences for affected developers.
Key Requirements
- •AI developers of potentially qualifying systems must be prepared to engage with the NSA-led classified benchmarking process used to designate 'covered frontier models'
- •Designated 'covered frontier models' are identified based on advanced cyber capability thresholds established through a classified assessment framework
- •Developers may voluntarily share frontier models with designated trusted government partners for security review under the established framework
- •No mandatory government licensing, permitting, or approval is required for the development or deployment of AI models under this order
- •Federal agencies are directed to coordinate interagency assessments of advanced AI cyber capabilities within the scope of the order
- •Developers operating in the frontier model space should monitor for formal designation notifications or agency outreach tied to the benchmarking process
What Your Organization Must Do
- →Audit all AI models in development or deployment to determine whether any may qualify as 'covered frontier models' based on advanced cyber capability profiles
- →Establish an internal classification and tracking process for frontier models to enable rapid response if NSA benchmarking outreach occurs
- →Designate a senior point of contact, such as a Chief AI Officer or Chief Information Security Officer, to manage any government security review engagement under the voluntary framework
- →Review and update model development and release policies to reflect the prohibition on mandatory licensing, ensuring legal and policy teams understand the scope of the voluntary framework
- →Engage legal counsel to assess confidentiality, classification, and intellectual property implications before voluntarily submitting any model to government partners for security review
- →Monitor agency guidance and Federal Register notices for implementing regulations, threshold definitions, or formal designation criteria as the classified benchmarking process matures