On June 2, 2026, President Donald J. Trump signed an executive order titled "Promoting Advanced Artificial Intelligence Innovation and Security." It was published in the Federal Register on June 5, 2026, as FR Doc. 2026-11415. If you run a business and you have felt one step behind on where AI policy is headed, this order gives you something concrete to read the direction from. It names the federal priorities in plain terms: cybersecurity, review of the most advanced AI models, protection of critical infrastructure, and criminal enforcement against people who misuse AI.
One thing is worth getting right up front. Most of the heavy obligations in this order fall on the largest frontier-model developers, not on the average company. For most leaders, the takeaway is direction, not a new pile of paperwork due next week.
The two things the order is trying to do
The order pursues two parallel approaches at once.
The first is defense. It directs the federal government and private industry to strengthen their cyber defenses against advanced AI. The second is review. It builds a voluntary benchmarking and review process for how the most capable AI models get developed and released.
That word "voluntary" is doing real work here. The order is explicit that nothing in it authorizes a mandatory governmental licensing, preclearance, or permitting requirement for the development, publication, release, or distribution of new AI models, including frontier models. There is no new permission slip a developer has to obtain before shipping. The review process is something developers can opt into, not a checkpoint they must clear.
According to the official fact sheet, the stated purpose is to advance American AI innovation in a way that strengthens cybersecurity, protects critical infrastructure, and keeps the United States as the global leader in AI. The fact sheet puts the posture this way: "The United States continues to lead the world in Artificial Intelligence (AI) because of the enormous talent and innovation of our AI industry, and because we refuse to stifle this innovation with overly burdensome regulation."
What a "covered frontier model" is
The order sets up a classified benchmarking process to assess the advanced cyber capabilities of AI models. That process is what sets the threshold at which a model gets designated a "covered frontier model." Think of it as the dividing line between an ordinary model and one capable enough on the cyber front to warrant a closer federal look.
The designation is not made by one person alone. The Director of the National Security Agency makes the determination, in consultation with the National Cyber Director, the Assistant to the President for Science and Technology, the Director of the Cybersecurity and Infrastructure Security Agency, and other representatives of the Department of War.
For developers building at that level, the voluntary framework offers three things they can choose to do:
- Engage the federal government to find out whether a model under development would meet the "covered frontier model" designation.
- Give the government access to a covered frontier model for a period of up to 30 days before they plan to release it to other trusted partners.
- Work with the government to select which trusted partners get that early access.
Again, all three are options a developer elects, not requirements imposed on them.
The deadlines that are already running
Several clocks started the day the order was signed.
Within 30 days, three things are due:
- The Secretary of Homeland Security, through the Director of CISA, must release Binding Operational Directives to speed up and prioritize the cyber defense of civilian federal information systems, set up or expand federal programs that use AI-enabled defensive tools, and ease access to cybersecurity tools and services (including, in some cases, covered frontier models) for agencies, state and local authorities, and operators of critical infrastructure.
- The Secretary of the Treasury, working with the National Cyber Director, the Secretary of War (through the Director of the NSA), and the Secretary of Homeland Security, must form an AI cybersecurity clearinghouse. In voluntary collaboration with the AI industry and critical-infrastructure operators, it coordinates scanning for software vulnerabilities and prioritizes patching and distribution of fixes.
- The Committee on National Security Systems must prioritize the cyber defense of National Security Systems.
The window for the CISA Binding Operational Directives and the Treasury-led AI cybersecurity clearinghouse. Both clocks started the day the order was signed.
Source: The White House, 2026Within 60 days, the Secretary of the Treasury, the Secretary of War (through the Director of the NSA), and the Secretary of Homeland Security (through the Director of CISA) must develop and maintain the classified benchmarking process that assesses AI cyber capabilities and sets the "covered frontier model" threshold.
The order names three categories of systems to prioritize for cyber defense: National Security Systems, Department of War information systems, and civilian federal government information systems.
"An AI did it" is not a defense
One section speaks directly to anyone deploying AI inside a business. The order directs the Attorney General to prioritize criminal enforcement against malicious AI-enabled cyber activity. It points to existing federal criminal statutes, including 18 U.S.C. 1028, 18 U.S.C. 1030, and 18 U.S.C. 1343, and applies them to anyone who uses AI to illegally access or damage a computer without authorization, or who uses AI in the course of committing another crime. The order specifically calls out employing AI agents to unlawfully access data later used for a criminal purpose.
The reading for a business leader is straightforward. If you are giving autonomous AI agents access to systems and data inside your company, you own what those agents do. Putting human oversight and real access controls around them is no longer just good hygiene. It is the difference between a tool you can account for and one you cannot.
If you are giving autonomous AI agents access to systems and data inside your company, you own what those agents do.
Where this connects to AI proficiency
This is the policy version of a point that runs through The 7 Levels of AI Proficiency: the value is not the model, it is the workflow and the human judgment built around it. A federal order grounded in voluntary review, defensive collaboration, and accountability for misuse rewards the same thing the higher levels of The 7 Levels of AI Proficiency describe, the ability to direct AI systems with oversight rather than just turn them loose.
A leader operating at those higher levels is not asking which model to install. They are asking who reviews the output, who holds the access keys, and what happens when an agent does something unexpected. The order makes that the national posture. Voluntary on the review side, serious on the accountability side, human in charge throughout.
A next step
Read the order or its official fact sheet yourself, and then look at your own AI agents through its lens. Where do they have access? Who is reviewing what they produce? Who is accountable if one of them does something it should not? You do not need to be a frontier-model developer for those questions to apply to you. Answering them is how a leader turns a federal order into a checklist for their own house.
Related reading: Level 7: Mission Director.
Sources
- Promoting Advanced Artificial Intelligence Innovation and Security (Presidential Action)
- Promoting Advanced Artificial Intelligence Innovation and Security (Federal Register, FR Doc. 2026-11415)
- Fact Sheet: President Donald J. Trump Promotes Advanced Artificial Intelligence Innovation and Security
Frequently Asked Questions
Does this order require my company to register or license our AI tools?
No. The order explicitly states it does not create a mandatory licensing, preclearance, or permitting requirement for developing, publishing, releasing, or distributing AI models, including frontier models. The review process is voluntary.
Who does the order actually obligate?
The action items fall on federal agencies, led by the Treasury, Homeland Security and CISA, the NSA, the National Cyber Director, and the Attorney General. The covered frontier model review is aimed at the developers of the most advanced models. Most businesses are observers of the direction, not subjects of new requirements.
What should a critical-infrastructure operator watch for?
The CISA Binding Operational Directives due within 30 days and the Treasury-led AI cybersecurity clearinghouse, also due within 30 days. Both are designed to expand access to federal cybersecurity tools for agencies, state and local authorities, and critical-infrastructure operators.
Is there a new criminal exposure here?
The order does not create new crimes. It directs the Attorney General to prioritize enforcement of existing federal criminal laws against malicious AI-enabled cyber activity, including the use of AI agents to unlawfully access data.
Find your AI Proficiency level
The free 7 Levels assessment places you across seven stages of AI capability. Under ten minutes. Research-backed scoring.