😬 Executive order disorder | AI+ Government Summary

"Executive Order Disorder" | May 22, 2026

1. Key Themes

Theme 1: The "Accelerationist" faction has consolidated control over U.S. AI policy — for now

The collapse of the AI executive order signals that the pro-growth, anti-regulation camp — led by tech executives and White House AI adviser David Sacks — has effectively vetoed safety-oriented AI governance. This creates a prolonged policy vacuum with real consequences for cybersecurity infrastructure.

"The whole thing was unnecessary and just something doomers wanted." — Source familiar with the White House deliberations

"For now, the accelerationists have won out." — Axios editors

Trump told reporters: "I think it gets in the way of — you know, we're leading China, we're leading everybody, and I didn't want to do anything to get in the way of that lead."

Theme 2: Open-source AI is gaining a concrete foothold in federal scientific infrastructure

Reflection AI's exclusive deal to serve as the "foundational intelligence layer" for the Department of Energy's 17 national laboratories is a landmark moment for open-weight models in government. It signals that the DOE's Genesis Mission is betting on customizability and transparency over proprietary systems — a meaningful procurement signal for the broader govtech AI market.

"You can't do scientific discovery on a closed model. It would be like doing rocket science, but you only get to look at the rocket, as opposed to taking it apart and examining the engine." — Reflection AI CEO Misha Laskin

"The company is aiming to be the premier U.S. AI lab for the most advanced open-weight models, where the trained parameters are available to the public."

Theme 3: The government's relationship with safety-focused AI labs is becoming legally and operationally incoherent

The Pentagon is simultaneously arguing in federal court that Anthropic is a supply-chain risk while actively evaluating Anthropic's newest model, Mythos, to combat cyberthreats. This contradiction represents a structural risk for any AI company with a stated safety mission that wants to do federal business.

"Treating a U.S. company as a national security threat while looking to use its technology to combat foreign adversaries is an awkward needle to thread." — Axios editors

D.C. federal appeals court Judge Karen Henderson: "To me this is just a spectacular overreach by the department."

Theme 4: AI content law is technically in effect but practically unenforceable — litigation will define implementation

The Take It Down Act passed its compliance deadline (May 19), but the absence of transparency requirements means platforms can implement removal processes inconsistently. Enforcement will likely be shaped reactively, through high-profile FTC lawsuits rather than proactive regulatory guidance.

"We might see a high-profile case and that might be how the FTC is choosing to provide that transparency in the coming months." — Omny Miranda Martone, CEO of the Sexual Violence Prevention Association

"Advocates say there is still a major transparency gap on how the decisions are being made to take down or leave up posts."

2. Contrarian Perspectives

The AI executive order's collapse may actually reflect industry winning — not dysfunction

The conventional read is that the EO's failure is chaos. The contrarian read: organized industry lobbying by Zuckerberg, Musk, and Sacks successfully killed a regulatory document days before signing. That's not disorder — it's a demonstration of how effective the anti-regulation coalition has become at shaping executive action.

"Meta CEO Mark Zuckerberg, xAI CEO Elon Musk and Sacks all spoke with Trump between Wednesday night and Thursday morning" — immediately before the signing was pulled.

"The main reason the signing was delayed was that Trump 'just hates regulation'" — suggesting the lobbying found a receptive ear with minimal friction.

The Pentagon's blacklisting of Anthropic may backfire strategically — the courts appear skeptical

The prevailing assumption is that national security designations are hard to challenge. But judicial pushback in this case was unusually sharp, with a sitting federal appeals judge calling it a "spectacular overreach." This suggests safety-focused AI companies have more legal leverage than widely assumed when fighting supply-chain risk designations.

Judge Karen Henderson: "For the life of me, I do not see any evidence of maliciousness despite the best efforts of [Pentagon Under Secretary Emil Michael], who in his memo refers to you as having mal-intent, a bad motive, cannot be trusted."

Open-weight models — not proprietary frontier models — may be the winning architecture for government AI

Conventional wisdom favors major closed-model providers (OpenAI, Anthropic) for high-stakes government work. But the DOE's choice of Reflection AI specifically because of open weights challenges that assumption, implying that for scientific and mission-critical use cases, customizability and auditability matter more than raw model capability.

"The reason this is the open model base of the Genesis mission is because scientists need full access to the model in order to be able to understand and customize it." — Misha Laskin

3. Companies Identified

Reflection AI

Description: Open-source/open-weight AI startup
Why mentioned: Exclusively announced as AI model provider for DOE's Genesis Mission, powering all 17 national laboratories
Key quote: "Reflection will provide DOE with AI models it can customize for its data, and Reflection will use DOE's compute as it is deployed across Genesis Mission research projects."

Anthropic

Description: AI safety-focused frontier AI lab, maker of the Claude model family
Why mentioned: Simultaneously blacklisted by the Pentagon as a supply-chain risk and being evaluated for its newest model (Mythos) for cybersecurity use — subject of active federal appeals court litigation
Key quote: "Anthropic argues it has no way to control its AI models once they're deployed in classified settings."

Meta

Description: Social media and AI conglomerate
Why mentioned: CEO Mark Zuckerberg directly lobbied Trump in the hours before the AI executive order was to be signed, contributing to its cancellation
Key quote: "Meta CEO Mark Zuckerberg...spoke with Trump between Wednesday night and Thursday morning."

xAI

Description: Elon Musk's AI company
Why mentioned: CEO Elon Musk was part of the last-minute lobbying effort that derailed the executive order
Key quote: "xAI CEO Elon Musk...spoke with Trump between Wednesday night and Thursday morning."

4. People Identified

David Sacks

Description: White House AI and crypto adviser to President Trump
Why mentioned: Key figure in killing the AI executive order; described as strongly opposed to the order's regulatory provisions
Key quote: "Sacks also 'hated it.'"

Misha Laskin

Description: CEO of Reflection AI
Why mentioned: Announced the DOE/Genesis Mission partnership; articulated the core argument for open-weight models in government scientific research
Key quote: "It's personal for me because I'm an immigrant to this country, and I moved here because my parents got jobs in a national lab."

Judge Karen Henderson

Description: D.C. Circuit Court of Appeals judge
Why mentioned: Delivered the most pointed judicial rebuke of the Pentagon's Anthropic blacklisting during oral arguments
Key quote: "To me this is just a spectacular overreach by the department."

Judge Gregory Katsas

Description: D.C. Circuit Court of Appeals judge
Why mentioned: Raised the fundamental challenge to Anthropic's position — that rapidly evolving AI models make usage policies difficult to enforce contractually
Key quote: "AI three months from now will be totally different from the AI of today."

Omny Miranda Martone

Description: Founder and CEO of the Sexual Violence Prevention Association
Why mentioned: Provided expert assessment of Take It Down Act implementation gaps and predicted litigation-driven enforcement
Key quote: "We might see a high-profile case and that might be how the FTC is choosing to provide that transparency in the coming months."

Emil Michael

Description: Pentagon Under Secretary
Why mentioned: Author of the memo designating Anthropic as a supply-chain risk; his language ("mal-intent," "bad motive") was directly criticized by the federal appeals court
Key quote: Referenced by Judge Henderson: "[Pentagon Under Secretary Emil Michael], who in his memo refers to you as having mal-intent, a bad motive, cannot be trusted."

5. Operating Insights

For govtech AI companies: Open-weight architecture may now be a procurement advantage, not just a philosophical stance

The DOE explicitly selected Reflection AI because scientists require full model access for customization and auditability. AI companies competing for scientific or mission-critical government contracts should evaluate whether open-weight offerings give them a differentiated path to procurement, particularly in an environment where closed-model providers face political and legal headwinds.

"Scientists need full access to the model in order to be able to understand and customize it." — Misha Laskin

For AI companies with government ambitions: A safety or ethics brand is now a legal and business liability in certain federal contexts — build contractual flexibility accordingly

Anthropic's stated AI safety policies have been weaponized against it in a Pentagon designation, with the argument that the company might "pull the plug" on deployed models due to "ideological" views. AI companies should structure their government contracts and usage policies to explicitly address deployment continuity in classified or critical settings.

"The Pentagon claims it's unworkable for the military or its vendors to rely on Anthropic because the company might pull the plug at any time due to its 'ideological' views around AI safety."

For compliance and legal teams: The Take It Down Act creates immediate enforcement risk even without FTC action — the mechanism is litigation, not rulemaking

Platforms cannot wait for regulatory clarity on the transparency gap; advocates are already expecting lawsuits. Companies should audit their content removal workflows now and document decision-making processes before a high-profile case sets binding precedent.

"Advocates say companies will have to be taken to court to secure strong implementation."

6. Overlooked Insights

The Treasury Department's proposed role in AI security vulnerability assessment is an anomalous — and potentially significant — institutional power grab

The article briefly flags that the now-delayed executive order would have assigned Treasury a leading role in identifying and fixing AI model security vulnerabilities — a function historically held by CISA and NIST. This quiet reassignment of cybersecurity authority to a financial regulator deserves scrutiny as the order is renegotiated; it could reshape which agency AI companies must engage for compliance.

"It's not clear just objectively speaking why Treasury is involved and what is their substantive expertise in this area." — Tech industry source

The FTC has already stood up infrastructure for Take It Down Act enforcement — before the courts weigh in

Buried in the deepfakes section is the fact that the FTC has launched a dedicated website for victims to report non-compliant platforms. This is an active enforcement posture, not a passive one — meaning the FTC is building a case pipeline even before litigation defines implementation standards.

"The Federal Trade Commission launched a website for victims to report when platforms have failed to act on valid requests."