Breaking

Lutnick Says Anthropic Is Back in Favor; Pentagon Official Says Supply-Chain Risk Label Still Stands

September 3, 2026 at 23:41 EDT

Policy & Regulation
Foundation Models
Sales & Business

U.S. Commerce Secretary Howard Lutnick said Anthropic has repaired its relationship with the Trump administration, but Under Secretary of Defense for Research and Engineering Emil Michael said the Pentagon’s supply-chain risk designation of the company remains in effect. The split, reported on September 3, 2026, leaves federal AI contracting in an unsettled state even after a court loss for the Defense Department.

September 2026 · U.S. government · Anthropic

Thaw at Commerce. Freeze at the Pentagon.

Howard Lutnick says Anthropic has done what the administration asked. Emil Michael says the Defense Department’s supply-chain risk designation of the Claude maker remains in effect.

Timeline of the clash

Feb 27 Trump orders agencies off Anthropic; 6-month phase-out.

→

March Pentagon supply-chain risk label; up to $200M contract at risk.

→

Aug 27 Judge Lin vacates the designation in a 59-page ruling.

→

Sept 3 Lutnick: resolved. Pentagon: label still on.

Pentagon contract

$200M

Ceiling of Anthropic’s defense deal before the February clash.

Phase-out window

6 mo

Time given agencies already using Claude to unwind deployments.

Court opinion

59 pp

Judge Rita Lin’s August 27 order vacating the blacklist.

Phase-out vs. immediate cutoff

Agencies already on Claude were given six months to unwind. New use was ordered to stop at once.

Immediate stop
for new use

Months to unwind
existing systems

Six stacked months of wind-down versus a one-block cutoff for new deployments.

Commerce view

Lutnick: Anthropic has done what was asked and is “back on the right side.” Tom Brown appeared at the G20 ministerial.

Pentagon view

Michael: The supply-chain risk designation remains in effect. A D.C. Circuit challenge is still pending.

What the court found

Judge Lin vacated the designation: retaliation, denied process, and no articulable sabotage risk.

Anthropic refused two uses: lethal autonomous weapons and mass surveillance of Americans.
Government conceded models cannot be altered or shut down once inside national-security systems.
Contractors still face split signals from Commerce and Defense.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.