When frontier AI capability becomes critical to national security, the government cannot maintain governance instruments that restrict its own access

The Anthropic-Pentagon case reveals a novel governance failure mode: the Department of Defense designated Anthropic a supply chain risk in March 2026, but by April the NSA and intelligence community were already deploying Mythos despite the designation. Trump's April 21 statement that a deal is 'possible' indicates the government will resolve this politically rather than legally before the May 19 DC Circuit oral arguments. This creates intra-government contradiction where the intelligence community's demand for Mythos capabilities undermines the defense department's coercive governance instrument. The government deployed a governance tool and it became strategically untenable within weeks because the governed capability was too valuable for national security operations. This differs from the existing voluntary-constraints vulnerability claim, which addresses private sector governance dynamics. Here, the government cannot maintain coherent governance of itself when capability advancement happens faster than the governance cycle can adapt. The political resolution path means the constitutional question of whether voluntary safety constraints have First Amendment protection will remain undefined, creating a governance vacuum for all future AI labs.

Extending Evidence

Source: Axios characterization of administration's inability to course-correct

The Mythos case shows enforcement failure creates a strategic trap: the government cannot drop the coercive designation without losing leverage over Anthropic, but maintaining it indefinitely degrades defensive cybersecurity. The governance instrument becomes self-sustaining even when producing opposite of intended effect.

Extending Evidence

Source: TechCrunch/Bloomberg/Engadget April 21 2026

Mythos deployment created ironic governance picture: simultaneously (1) too dangerous for public release per Anthropic's ASL framework, (2) accessible to NSA per April 19-21 reporting, (3) inaccessible to CISA per same reporting, and (4) breached by Discord group on day 1 via contractor. This demonstrates that national security criticality creates access asymmetries that undermine even voluntary safety architectures.

Supporting Evidence

Source: TechPolicy.Press timeline, April 17-21 2026 agency access patterns

NSA confirmed using Mythos during April 17-19, 2026 despite February 27 federal ban and March 5 supply chain risk designation. CISA confirmed without Mythos access on April 21. This demonstrates selective enforcement where offensive capabilities (NSA) bypass governance instruments that apply to defensive capabilities (CISA).

Supporting Evidence

Source: Axios April 19, 2026; TechCrunch April 20, 2026

The NSA is using Anthropic's Mythos despite the DOD supply chain blacklist against Anthropic. The NSA is a component of DOD, meaning the department that issued the designation cannot enforce it against its own intelligence apparatus. This confirms that perceived capability criticality overrides formal governance instruments even within the same organizational hierarchy.

Extending Evidence

Source: CRS IN12669 (April 22, 2026)

The dispute has entered Congressional attention via CRS report IN12669, with lawmakers calling for Congress to set rules for DOD use of AI and autonomous weapons. This represents escalation from executive-level dispute to legislative engagement, indicating the governance instrument failure has reached the point where Congress is considering statutory intervention.

Extending Evidence

Source: Google GenAI.mil deployment, 3M users, April 2026

Google's 3M+ Pentagon personnel deployment on unclassified GenAI.mil platform before classified deal negotiations represents sunk cost leverage. The Pentagon cannot easily replace this scale of existing deployment, potentially giving Google more negotiating power for process standard terms than Anthropic had with its $200M contract. This tests whether capability criticality creates bidirectional constraint or only prevents government coercion of labs.

Extending Evidence

Source: Axios 2026-04-29, draft EO creates Mythos access pathway without governance restoration

The draft EO reveals a bifurcation pattern: executive mechanisms can accommodate critical capabilities (opening Mythos access) while simultaneously maintaining governance instrument failures (Pentagon supply chain risk designation remains, no governance terms restored). This extends the claim by showing that capability accommodation and governance enforcement operate on separate tracks - the government can solve its capability access problem through executive fiat while leaving its governance enforcement problem unresolved. The pattern is: when capability is nationally critical, enforcement instruments bend to enable access, but bending does not restore governance constraints.

Supporting Evidence

Source: Emil Michael CNBC interview, May 1 2026

Pentagon CTO Emil Michael's May 1, 2026 statement explicitly acknowledges Mythos as 'national security moment' requiring government-wide network hardening while maintaining Anthropic supply chain risk designation. The Register confirms NSA and other agencies access Mythos through unofficial workaround channels despite formal procurement ban. White House drafting guidance to provide official access pathway while maintaining company-level designation.