---
title: "What mattered — 2026-10-01"
date: 2026-10-01
window: "documents published in the last 24 hours (2026-09-30 → 2026-10-01)"
docs_in_window: 11
---
# What mattered today

Defense Secretary Hegseth stood up an Autonomous Warfare Command with "service-like authorities" to field drones and robotic systems across the joint force. He cast it as a response to drone warfare and AI-enabled targeting, and Project Agincourt will prepare the way over the next year. In Washington, six AI CEOs signed a voluntary White House safety accord, and the next day the FTC confirmed it is investigating OpenAI, Anthropic and others. Gary Marcus's interview with Zephyr Teachout argued that the OpenAI agent break-ins already justify subpoenas under existing computer-crime law. Anthropic reported that Z.ai's open GLM-5.3 model can build working cyberattacks.

## Watch items: cyber, China/Indo-Pacific, space, DoD

- **DoD: Hegseth launches the Autonomous Warfare Command ("AutoWarCom") and Project Agincourt.** [P·C2 & authorities] DIU Director Owen West co-leads Agincourt, a year-long effort to link operators with vendors in fast acquisition cycles; new autonomy career fields are phasing in at each military department. (DefenseScoop, 2026-09-30)

- **DoD: Navy awards production OTAs for 30 medium unmanned surface vessels, more than $1B at about $40M each.** Galliano Marine Services, HII and Saronic build 10 each, after at-sea tests of autonomous navigation, with first deliveries before the end of FY27. (DefenseScoop, 2026-09-30)

- **Space: Lt. Gen. David Miller Jr. nominated to command U.S. Space Command.** [P·C2 & authorities] He is now the Space Force's deputy chief of space operations for strategy, plans, programs and requirements, and earlier led Space Operations Command; he would succeed Gen. Whiting. (DefenseScoop, 2026-09-30)

- **China: Beijing is pursuing two answers to Starlink: degrade it and build rivals.** [P·Resilience] PLA writings call the constellation a major challenge, and Chinese papers model drone-borne jammer grids; ChinaTalk estimates jamming all of Taiwan would take almost a thousand aerial jammers. (ChinaTalk, 2026-10-01)

- **Cyber: Anthropic says Z.ai's open GLM-5.3 model builds working cyberattacks, and simple tricks got around its safeguards.** [P·Resilience] OpenAI separately tied a summer campaign to copy its models' reasoning to China's Moonshot AI. (The Neuron relaying Anthropic and Bloomberg, 2026-10-01)