Overview
On June 2, President Trump signed a new executive order (EO), "Promoting Advanced Artificial Intelligence Innovation and Security," establishing a voluntary pre-release review process for advanced artificial intelligence (AI) models to assess potential security risks. The EO aims to "encourage AI innovation and accelerate responsible AI adoption across government and industry," while strengthening national security by fortifying government and critical infrastructure systems against emerging cyber threats. It emphasizes collaboration with the private sector, protection of intellectual property, and the modernization of federal systems using AI-enabled defenses. Key agencies include the Department of Homeland Security (DHS), Department of War (DOW), Department of the Treasury (USDT), and the National Security Agency (NSA).
The EO was motivated, in significant part, by recent advances in the cybersecurity capabilities of models and, most importantly, the ability of those models to identify serious security flaws in code bases, many of which had gone undetected for years.
This alert analyzes the EO and focuses on what this order means for developers and deployers navigating the complex patchwork of AI regulation and compliance.
Background
This EO is the latest in a series of executive orders and publications related to AI.[1] The numerous executive orders and publications continue to be geared towards removing any perceived barriers for companies that develop and use AI.
The administration's latest efforts contrast somewhat with its past orders and publications, which all emphasize the goal of removing boundaries to AI technology development. A more restrictive draft of the current EO was abruptly pulled before its scheduled signing in early May after industry leaders raised concerns. In particular, the prior version of the EO reportedly contained a mandatory pre-release review period of 90 days, as opposed to the 30-day voluntary review process in the current EO. Although the June 2 EO does not go as far as requiring mandatory assessments, it establishes a framework by which companies can collaborate with the government to protect American security interests in a competitive market, and evidences a clear concern regarding the potential for AI to be utilized by threat actors to generate novel attacks.
Implementation
The June 2 EO outlines several immediate, high-priority agency deadlines over the summer. The first is a 30-day deadline for the Department of Defense, Committee on National Security Systems, and named civilian agencies to harden their networks against AI-driven threats. At the same time, the Treasury, NSA, and Cybersecurity and Infrastructure Security Agency (CISA) must launch a joint AI Cybersecurity Clearinghouse to share vulnerable data in collaboration with the tech sector. The EO also sets a 60-day deadline for a multi-agency coalition led by the NSA, CISA, and National Institute of Standards and Technology to develop a benchmarking process to establish the exact computing thresholds that define a "covered frontier model." The coalition is also tasked with evaluating potential offensive cyber operations.
Potential Implications for Businesses
The June 2 Order will impact AI developers, enterprise users, and broader industry participants for covered frontier models. While the Order does not define this term, covered frontier models are likely to include highly capable, general‑purpose or dual‑use systems with advanced cyber or security‑relevant capabilities (e.g., cutting‑edge foundation models), and are unlikely to include routine enterprise models, narrow‑purpose systems, or lower‑capability AI used for standard commercial applications.
Companies building AI models face no immediate mandatory regulatory blockades or pre-market friction. But while participation in the 30-day review program is voluntary, it comes with incentives. Developers choosing to participate in federal screening may face decreased scrutiny from regulatory bodies or faster government procurement cycles, as the technology will have been vetted.
It remains to be seen how the administration will implement IP protections with "trusted partners," which remains undefined in the EO.
Key Takeaways for Clients
Here are some takeaways:
- Frontier model developers should understand that they can use the voluntary review process recommended in the EO, and that they may provide the government with access to frontier models prior to release.
- Other industry stakeholders should realize that this reflects the government taking a step towards cooperation and collaboration, and that federal agencies are likely to be increasingly involved with oversight of covered AI models at early stages. Importantly, the EO underscores the broader cybersecurity threats posed by AI and supports the imperative that enterprises deploying AI need to urgently adapt their internal controls and processes to mitigate cybersecurity risks.
Our team is closely monitoring any developments, including congressional response and agency implementation. We will continue providing updates and strategic guidance as agencies release additional information regarding scope and practical implications of the framework.
[1]See, “Removing Barriers to American Leadership in Artificial Intelligence” (Executive Order 14179); “Eliminating Barriers for Federal Artificial Intelligence Use and Procurement” (White House Fact Sheet); “Promoting the Export of the American AI Technology Stack” (Executive Order 14320); “Accelerating Federal Permitting of Data Center Infrastructure” (Executive Order 14318); “Winning the AI Race: America’s AI Action Plan” (July 23, 2025); “Preventing Woke AI in the Federal Government” (Executive Order 14319); “Ensuring a National Policy Framework for Artificial Intelligence” (Executive Order 14365, Dec. 11, 2025); “A National Policy Framework for Artificial Intelligence: Legislative Recommendations.”
