EO 14409: voluntary pre-release review of frontier models
Confirmed
Importance: major (4 of 5)The takeaway
On June 2, 2026 President Trump signed Executive Order 14409, “Promoting Advanced Artificial Intelligence Innovation and Security”.
Status
- Claim
Confirmed
- Our reporting
- High confidence
- Importance
- Major (4 of 5)
- Last verified
- 14 October 2026
Your AI and this story
- GPT-6 Astra33 days after its cutoff
- Claude Opus 5.5In its training data
- Gemini 4 Argon63 days after its cutoff
- Grok 4.72 days after its cutoff
It is after the cutoff of GPT-6 Astra, Gemini 4 Argon and Grok 4.7, and can be in the training data of Claude Opus 5.5.
Key facts
- Executive Order 14409, signed June 2, 2026 (whitehouse.gov)
- Sec. 3 (60 days): a classified benchmarking process for advanced cyber capabilities sets the ‘covered frontier model’ threshold; the NSA Director decides, consulting the National Cyber Director, the science adviser (APST), CISA and others. The term is not defined in the text
- Voluntary framework: developers give the government access for up to 30 days before release to other trusted partners, with confidentiality, cybersecurity, insider-risk and IP protections
- Sec. 3(c): ‘Nothing in this section shall be construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement’ for new AI models, including frontier models
- 30-day tasks: cyber defense of National Security Systems, Department of War systems and civilian federal systems (CISA directives, which also widen access to defensive AI tools, including covered frontier models, for state and local governments and critical-infrastructure operators); a voluntary Treasury-led AI cybersecurity clearinghouse with industry for vulnerability scanning, validation and patching; OMB review of grant funding for AI vulnerability detection
- 60 days: OPM to expand cyber hiring pathways under the US Tech Force. Sec. 4: the Attorney General is to prioritize prosecuting AI-enabled computer crime
- CAISI is not mentioned; NIST appears once, as a consulted agency
- Trump pulled an earlier draft in May, reportedly partly over its longer (90-day) review window (Infosecurity Magazine, Gizmodo)
Show 2 more
- Reactions: Anthropic called it ‘an important step in strengthening America’s leadership in AI’; OpenAI called it ‘an important step forward’ and in a separate paper asked for binding rules including mandatory CAISI pre-release testing (The Decoder)
- Widely read as a response to cyber-capable models such as Anthropic’s Claude Mythos Preview; the framework resembles Anthropic’s Project Glasswing early-access program (Infosecurity Magazine; the order names no model)
What happened
After months of debate over how the US government should see the most capable models before release, the White House chose a voluntary, cyber-focused route. The NSA, not the Commerce Department’s CAISI, leads the classified test that decides which models count as “covered”. Developers who join give the government up to 30 days of early access. The rest of the order is about defense: hardening federal systems, spreading AI defensive tools to smaller operators and a vulnerability clearinghouse run with industry.
Why it matters
It is the main US federal rule for pre-release access to frontier models in 2026 and shapes what followed. Three weeks later the administration asked OpenAI to limit GPT-5.6’s first release to approved partners (2026-06-25 entry). The WSJ reported on June 10 that CAISI was told to stop publishing model assessments while the order was implemented, and in October the White House moved CAISI officials into its own staff. By ruling out licensing, it also marks the line later fights in Congress and the states argue over.
Sources
4 sources from 4 sites. Numbers match the chips in the text.
4 sources: 1 primary, 3 press
Primary
- White House: Executive Order 14409, Promoting Advanced Artificial Intelligence Innovation and Security (June 2, 2026)whitehouse.gov, official
Press
- Infosecurity Magazine: Trump signs order inviting voluntary review of frontier AI models (June 3)infosecurity-magazine.com, press
- The Decoder: Trump’s new executive order wants AI companies to voluntarily submit models for government safety reviews (June 3)the-decoder.com, press
- eSecurity Planet: Trump signs executive order creating voluntary AI security review frameworkesecurityplanet.com, press
Changes
- Filed (found while writing The Information: the Trump administration plans to move three senior officials of the US AI testing agency (CAISI) to the White House); order text read on whitehouse.gov