ONYX covered the Pentagon’s ‘supply chain risk’ designation against Anthropic in the August arc — a designation that effectively blocked Anthropic from US government contracts. ONYX covered Amodei’s ‘We Must Pace the Frontier’ essay (September 13-14), his CBS Sunday Morning interview where he said he didn’t fully anticipate AI’s rapid effects, and the AI safety industry-Congress debate. Today: the Pentagon has rescinded the designation.
THE REVERSAL
The US Defense Department rescinded its ‘supply chain risk’ designation against Anthropic, the AI safety company that makes Claude, according to Reuters and NPR. The specific consequence of the reversal: Anthropic can now be a vendor for US government contracts. The designation had effectively excluded Anthropic from federal procurement.
WHAT CHANGED
The specific reason for the reversal is from developing reporting. The timing: the reversal comes in the same week as Amodei’s public AI safety essay generating significant attention, the congressional debate on AI regulation, and the Pentagon’s ongoing AI governance work. Whether the reversal reflects a changed security assessment, a policy decision to engage Anthropic in government AI work, or a response to the AI safety debate’s political momentum is from the specific reporting on the Pentagon’s stated rationale.
THE SIGNIFICANCE
The ‘supply chain risk’ designation is a specific national security designation that the Defense Department applies when it assesses a company poses risks in the government’s supply chain. Applying it to an AI safety company — one whose specific mission is reducing AI risk — produced documented tension between the Pentagon’s AI risk assessment and Anthropic’s self-description as an AI safety organization. The rescission resolves that documented institutional tension.
THE AI SAFETY DEBATE CONTEXT
The Pentagon reversal follows the week in which Amodei proposed METR embedded evaluators at every frontier AI lab, Altman and Musk endorsed the proposal, Congress debated AI regulation, and Speaker Johnson urged against panic. Whether the Pentagon’s reversal is connected to Amodei’s public safety engagement is not confirmed by reporting on the specific timing and rationale. ONYX notes the temporal proximity without asserting causation.
The Pentagon called Anthropic a supply chain risk. Then Anthropic’s CEO went on CBS, published a 3,400-word AI safety essay, and called for embedded third-party evaluators at every frontier AI lab. This week, the Pentagon reversed the designation. Both things happened. The connection between them is in the Pentagon’s stated rationale, which ONYX will cover when confirmed.
WHAT HAPPENS NEXT
▸ Pentagon rationale — the specific stated reason for the reversal
▸ Government contracts — whether Anthropic pursues specific federal procurement opportunities
▸ AI governance — whether the reversal affects the broader AI safety-government relationship
| CONFIDENCE: HIGH | Pentagon rescinded the supply chain risk designation for Anthropic, an AI safety company and maker; Reuters and NPR confirmed the reporting. |
SOURCES
▸ Reuters / NPR — Pentagon Anthropic supply chain risk rescinded September 18, 2026
