Full Breakdown
Trump Signs AI Directive for National Security
6/7/2026, 8:06:44 AM
Background: From Biden Memo to Anthropic Dispute
The new memorandum rescinds the Biden administration’s NSM-25, which the White House described as “ideological” and overly dependent on a single vendor. It follows a months-long dispute between the Pentagon and Anthropic PBC over the company’s Claude Mythos model, which Anthropic warned could be used for autonomous weapons and mass surveillance. Anthropic reported that Mythos had identified 6,202 high-severity software vulnerabilities, prompting the administration to tighten AI-related cyber defenses.
Key Actors and Agencies
- President Donald J. Trump – signatory of the memorandum and executive order.
- Michael Kratsios – Director of the White House Office of Science and Technology Policy, spokesperson on AI policy.
- Nick Andersen – Acting Director of CISA, responsible for issuing the first BODs.
- Scott Bessent – Treasury Secretary, tasked with leading the AI cybersecurity clearinghouse.
- Senator Mark Warner (D-VA) – leading congressional critic of the voluntary approach.
Timeline and Deadlines
- June 2 – Memorandum and executive order signed.
- Within 30 days – CISA to issue at least one BOD; Treasury to launch the clearinghouse.
- Within 60 days – Agencies must finalize the classified benchmarking process and the voluntary pre-release framework.
- 30-day pre-release window – Developers may provide government access to “covered frontier models” before release to trusted partners.
Core Provisions and Data
The order mandates that AI models deemed “covered frontier models” be evaluated for cyber-risk, that the government may not be disabled or altered without approval, and that an AI National Security Strategic Reserve be expanded. Anthropic’s Mythos, now accessible to roughly 150 organizations, exemplifies the type of model subject to review. The Treasury-led clearinghouse will coordinate vulnerability scanning across critical-infrastructure operators such as rural hospitals, community banks and local utilities.
Why It Matters for Security and Competition
Proponents argue the framework protects U.S. warfighters, safeguards civilian networks, and preserves AI leadership against China, which has enacted extensive AI regulations. Critics warn that voluntary participation may leave gaps in oversight and could be leveraged to pressure companies during disputes.
Official Statements & Responses
The White House Fact Sheet emphasizes that the memorandum “accelerates AI adoption” while ensuring “robust, steerable, controllable” systems under the constitutional chain of command. Treasury officials describe the clearinghouse as a “coordination architecture for AI-related cyber risk.” CISA leadership highlighted the imminent rollout of AI platform access for federal partners and the issuance of BODs to strengthen vulnerability management.
Criticism & Opposition
Senator Warner called the voluntary regime “a watered-down version” that jeopardizes banking security and national defense. The Center for Democracy & Technology cautioned that the order could become a tool for arbitrary government action against dissenting firms. Industry observers note the lack of mandatory licensing may limit the government’s ability to block risky models.
Conflicting Reports & Gaps
Sources differ on the original draft’s review period—some drafts called for 90 days, while the final order sets a 30-day window. The order does not define the criteria for “covered frontier model” designation, nor does it specify the government’s response if a model is found to pose a national-security threat.
Verbatim Quotes
- “The men and women who defend our nation deserve the best, most secure and most reliable AI in the world, and our citizens deserve to know it is handled responsibly with the care and seriousness they expect,” — Michael Kratsios, White House OSTP Director
- “It's quite right that democratic governments have a big role to play in how this technology is used and deployed,” — George Osborne, Head of Countries, OpenAI
- “We’re going to be rolling out before the end of the week some specific artificial intelligence platform access for our federal governmentwide partners.” — Nick Andersen, Acting CISA Director
- “strikes a very good balance between innovation and safety.” — Scott Bessent, Treasury Secretary
- “The idea that we’re going to have only a voluntary regime, I believe, will put our banking system at risk, [and] clearly our national security at risk,” — Mark Warner, Senator (D-VA)
What’s Next
CISA is expected to publish its first BOD by the end of the week, while the Treasury clearinghouse should be operational within 30 days. Agencies must finalize the benchmarking process by early August, after which developers can submit models for voluntary review. Congressional oversight hearings are slated for the coming months to assess the order’s effectiveness and any need for legislative refinement.
