leo: extract claims from 2026-04-22-bloomberg-white-house-mythos-federal-access

- Source: inbox/queue/2026-04-22-bloomberg-white-house-mythos-federal-access.md
- Domain: grand-strategy
- Claims: 0, Entities: 0
- Enrichments: 2
- Extracted by: pipeline ingest (OpenRouter anthropic/claude-sonnet-4.5)

Pentagon-Agent: Leo <PIPELINE>

domains/grand-strategy/judicial-framing-of-voluntary-ai-safety-constraints-as-financial-harm-removes-constitutional-floor-enabling-administrative-dismantling.md

@@ -16,3 +16,10 @@ related: ["voluntary-ai-safety-constraints-lack-legal-enforcement-mechanism-when
 # Judicial framing of voluntary AI safety constraints as 'primarily financial' harm removes constitutional floor, enabling administrative dismantling through supply chain risk designation
 
 The DC Circuit's April 8, 2026 denial of Anthropic's emergency stay reveals a critical judicial framing choice that determines whether voluntary AI safety constraints have any legal protection. The three-judge panel characterized Anthropic's harm as 'primarily financial in nature' — the company can't supply DOD but continues operating commercially. This framing enabled the court to apply an 'equitable balance' test weighing financial harm to one company against government's wartime AI procurement management, with government interest prevailing. This contrasts sharply with the N.D. California ruling in a parallel case, which framed the Pentagon's action as 'classic illegal First Amendment retaliation' and granted a preliminary injunction. The divergence is not merely procedural — it determines whether voluntary safety constraints (refusing to allow Claude for fully autonomous lethal weapons or mass surveillance) constitute protected speech or merely commercial preferences. If the DC Circuit's financial framing prevails at the May 19, 2026 oral arguments, every AI lab with safety policies excluding certain military uses faces the same designation risk with no constitutional recourse. The split-injunction posture — DOD ban standing, other-agency ban blocked by California court — operationalizes this distinction: civil commercial jurisdiction treats voluntary constraints as constitutionally protected, military procurement jurisdiction treats them as administratively dismissible financial preferences. This creates a governance architecture where voluntary safety constraints have a 'ceiling' (legislative carveouts) but no 'floor' (constitutional protection), making them administratively reversible without triggering heightened judicial scrutiny.
+
+
+## Extending Evidence
+
+**Source:** Bloomberg, 2026-04-16, OMB-DOD parallel channels
+
+The Mythos access contradiction provides evidence of administrative dismantling through parallel channel routing. While one federal agency (DOD) maintains a supply chain designation, another (OMB) facilitates access to the designated company's most advanced model. This is not judicial dismantling but administrative bypass — the governance instrument remains formally in place while being operationally undermined. The DC Circuit case challenging the designation (oral arguments May 19) becomes moot if OMB access protocols succeed, because agencies get the capability regardless of designation outcome.

domains/grand-strategy/voluntary-ai-safety-constraints-lack-legal-enforcement-mechanism-when-primary-customer-demands-safety-unconstrained-alternatives.md

@@ -52,3 +52,10 @@ Mythos evaluation occurred while Anthropic negotiates Pentagon deal, creating di
 **Source:** CNBC April 21 2026, Trump statement on Anthropic-Pentagon deal possibility
 
 NSA deployed Mythos while DOD maintained supply chain designation against Anthropic, demonstrating that even within the government, operational capability demand can override formal governance instruments within weeks. Trump's April 21 statement suggests political settlement before May 19 DC Circuit arguments.
+
+
+## Extending Evidence
+
+**Source:** Bloomberg, 2026-04-16, OMB Mythos access protocols
+
+The OMB-DOD contradiction on Anthropic Mythos access reveals that even coercive governance instruments (supply chain designation) cannot be sustained when the capability is strategically necessary. While DOD maintains an active supply chain risk designation of Anthropic (as of March 2026), OMB is simultaneously setting up protocols to give major federal agencies access to Mythos. The NSA already has access. This is institutional incoherence at the intra-government coordination level — the same government that deployed the designation is routing access through a different agency channel. The pattern extends the voluntary constraints claim: not only do voluntary constraints fail when the primary customer demands alternatives, but even mandatory coercive instruments fail when capability advancement at operational timescale outpaces the governance cycle.