What happened
On September 3, 2026, Under Secretary of Defense Emil Michael publicly stated that Anthropic 'is still a designated Supply Chain Risk at @DeptofWar and for the Defense Industrial Base,' one day after Commerce Secretary Howard Lutnick told Bloomberg/Axios that Anthropic had 'gotten religion' and repaired relations with the administration. This is a new development beyond the Aug 27 district court ruling (already covered) that vacated the Pentagon's designation under 10 U.S.C. §3252: Michael's statement concerns a separate FASCSA designation under 41 U.S.C. §4713, which remains legally operative pending a still-undecided D.C. Circuit appeal, and represents the Pentagon publicly signaling it will not back down despite the injunction and despite Commerce's rapprochement.
Why it matters
This creates active legal and commercial uncertainty for a major frontier AI lab and every defense contractor that might do business with it — two federal agencies are sending contradictory signals about whether Anthropic is bannable as a national-security risk. It shows the Trump administration's AI national-security posture is genuinely split between agencies, with real consequences for defense-sector AI procurement and Anthropic's pending IPO plans.
Action needed
Defense contractors and agencies using Anthropic's Claude/Mythos models should confirm current FASCSA compliance status pending the D.C. Circuit ruling; monitor for further statements reconciling the Commerce/Defense split.