Breaking

Pentagon Seeks to Lift Anthropic's Autonomous-Weapons Limits, Sparking Contract Dispute

June 23, 2026 at 06:14 EDT

Policy & Regulation
AI Agents
Foundation Models

The U.S. Department of War asked Anthropic to lift restrictions on using Claude for fully autonomous weapons and mass domestic surveillance, and after being refused, the matter has escalated into a legal dispute over an estimated $200 million military contract. As AI deployment on the battlefield accelerates with the Ukraine war serving as a testing ground, a clash over usage policy between the U.S. government and a private AI company has come to the surface.

National Security · AI Warfare

AI Moves Into the "Kill Chain" — and the Pentagon and Anthropic Collide

As frontier AI shifts from image analysis into targeting and decision loops, a dispute over autonomous weapons and surveillance has split Washington from one of its most-used AI suppliers.

$200M

Contract ceiling, each, to Anthropic · Google · OpenAI · xAI (July 2025)

+$795M

Ceiling increase for Palantir's Maven Smart System

30 days

Voluntary pre-release review for national-security models (EO, June 2026)

Federal AI funding by program

Approximate dollar scale, drawn to proportion

$795M

Maven Smart System
(Palantir increase)

$200M

Per-company ceiling
(4 AI firms)

The Anthropic Dispute

Anthropic refused unrestricted access for mass domestic surveillance and fully autonomous weapons — citing the reliability limits of today's frontier models.

Pentagon response

Agencies directed to stop using Anthropic; company designated a supply-chain risk.

The irony

Claude remains the Department's most widely deployed frontier AI — analysis, planning, cyber.

AI's path into the targeting loop

2017 — Project Maven
image recognition

→

Ukraine drones
AI for survival

→

Targeting & assessment
accelerated decisions

The case for guardrails

Frontier models are "simply not reliable enough" for fully autonomous weapons today, says Dario Amodei — and AI demands entirely new risk-analysis approaches.

The case for speed

Critics warn that refusal endangers national security, pointing to the speed-and-scale advantage of AI-driven drone swarms on the battlefield.

Continue reading

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