AI Executive Order Explained: The June 2, 2026 Trump AI Order
A clear, plain-English summary of the new federal AI executive order — what it does, who it covers, and what your business should do next.
The AI executive order explained here is the order President Trump signed on June 2, 2026. Its official title is "Promoting Advanced Artificial Intelligence Innovation and Security." It is the biggest federal AI policy action since President Biden's EO 14110 in 2023.
The new order is voluntary, not mandatory. It asks frontier AI developers to give the government a 30-day look at their most powerful models before public release. It also sets up new federal cyber-defense work focused on AI.
This AI executive order explained guide walks through every key provision in plain English. You will learn who it covers, what the deadlines are, how it compares to the Biden order, and what to do if your company uses or builds AI.
What the AI executive order says, in plain English
The order makes one big policy choice: voluntary review over mandatory rules. Companies are asked to opt in. The government will not require licenses or permits for new AI models.
The order also creates new federal AI cyber-defense work. Agencies must build AI cyber-capability benchmarks and a clearinghouse to share AI vulnerabilities.
It targets a narrow group — "covered frontier models." Most small businesses, schools, and developers are not directly affected by the new rules.
- Voluntary 30-day pre-release review of "covered frontier models" by the federal government.
- New federal AI cybersecurity clearinghouse to share vulnerability information.
- New AI cyber-capability benchmarks built by federal agencies to flag risky models.
- Explicit ban on mandatory federal AI model licensing or permits.
- Federal agencies told to harden their own AI security defenses.
- "Trusted partners" framework so private testers can also see models early.
Who the AI executive order applies to
The voluntary review piece targets frontier AI developers — the small handful of labs that train the world's most powerful models. OpenAI, Anthropic, Google DeepMind, Meta, and xAI are the obvious candidates.
The cyber-defense piece touches every federal agency. Federal contractors that handle AI for the government will also see new security rules.
Most companies, schools, and small developers are not directly covered. But the order will shape the broader market — including the AI tools you buy.
- Frontier AI labs — voluntary 30-day review and "covered model" benchmarking.
- Federal agencies — new internal AI security standards and cyber-defense work.
- Federal contractors — new AI security requirements when delivering AI to the government.
- Enterprise AI buyers — indirect impact through vendor benchmarks and the cyber clearinghouse.
- Small businesses and schools — no direct obligations under this order.
How it compares to the Biden AI executive order (EO 14110)
President Biden signed EO 14110 on October 30, 2023. It was the broadest federal AI policy action up to that point. President Trump rescinded most of it in January 2025.
The new June 2, 2026 order keeps the idea of frontier-model review. But it makes review voluntary, drops Defense Production Act reporting, and removes most of Biden's civil-rights and labor provisions.
For companies, the big change is less paperwork and fewer federal mandates. For frontier labs, the rules are now opt-in rather than required.
- Biden EO 14110 used the Defense Production Act to require frontier-model reporting. The new EO is voluntary.
- Biden's order included civil-rights, labor, and consumer-protection AI provisions. Most are removed.
- Biden's order created the US AI Safety Institute (US AISI). The Trump administration rebranded it as CAISI (Center for AI Standards and Innovation).
- Both orders ask for AI cyber-defense work — but the new order leans harder on private partnerships.
- Neither order preempts state AI laws. Colorado, California, Texas, and Utah laws still apply.
What the AI executive order does not do
A lot of news headlines suggested the order was a sweeping AI regulation. It is not. The order is narrow on purpose.
The White House explicitly framed the order as innovation-first. The text bars mandatory licensing and leaves most AI policy to Congress and the states.
For most businesses, the right read is "no new direct obligations — but the AI vendor landscape will shift."
- Does not create a mandatory AI model license or permit.
- Does not preempt state AI laws (Colorado, California, Texas, Utah all still apply).
- Does not impose new rules on AI use in schools, hiring, or healthcare.
- Does not require AI labeling for consumer products.
- Does not change federal copyright or IP rules for AI training data.
How to prepare for the AI executive order
Most companies do not need a new compliance program just for this order. But you should update three things: your vendor due-diligence, your AI inventory, and your federal-contract readiness.
If you sell AI to the government or use AI on government work, the cyber and benchmarking rules will hit your contracts within 12 months. Start now.
- Update your AI vendor checklist — add a question about voluntary federal review participation.
- Ask frontier-model vendors for their "covered model" status and benchmark results.
- Refresh your AI inventory — track which models touch federal data or contracts.
- If you are a federal contractor, watch for new AI cyber requirements in FAR clauses.
- Continue following NIST AI RMF — the EO reinforces it as the federal baseline.
- Keep state-law compliance plans in place. The new EO does not change Colorado, California, or Texas requirements.
- Re-check the AI law compliance tracker quarterly — implementation guidance will land throughout 2026.
Frequently Asked Questions
- The June 2, 2026 AI executive order is a federal policy that asks the most powerful AI developers to share their models with the government 30 days before public release. It is voluntary. It also creates new federal AI cyber-defense programs but does not require any AI license or permit.
- Only frontier AI developers are directly asked to participate in the voluntary 30-day review. Federal agencies and federal contractors face new AI security rules. Most other companies — including small businesses, schools, and developers — have no direct obligations under this order.
- The order took effect when President Trump signed it on June 2, 2026. Specific federal-agency benchmarks, the AI cybersecurity clearinghouse, and trusted-partner programs roll out over the following 6–12 months. Expect implementation guidance throughout the second half of 2026.
- It replaces what is left of EO 14110. President Trump rescinded most of Biden's order in January 2025. The new June 2, 2026 order keeps the idea of frontier-model review but makes it voluntary, drops Defense Production Act reporting, and removes Biden's civil-rights and labor AI provisions.
- No. The order does not preempt state AI laws. Colorado SB 24-205, Texas TRAIGA, California SB 53, California AB 2013, Utah SB 149, and Illinois HB 3773 all still apply to companies operating in those states.
- The order asks federal agencies to build benchmarks that define which AI models count as "covered frontier models" based on advanced cyber capabilities. The exact threshold will be set during implementation. Expect it to align loosely with California SB 53 (10^26 FLOPs) and the EU AI Act systemic-risk tier (10^25 FLOPs).
- Not directly. The order targets frontier AI labs and federal contractors. Small businesses have no new obligations under it. But your AI vendors may change their release timelines and disclosures — keep your vendor due-diligence checklist current.
Get a 30-minute briefing on what the new AI executive order means for you
Layer3 Labs gives you a clear, plain-English read on how the June 2, 2026 AI executive order affects your business — and a short list of the only changes you actually need to make.
Book your free AI EO briefing