---
title: "What mattered — 2026-09-15"
date: 2026-09-15
window: "documents published in the last 24 hours (2026-09-14 → 2026-09-15)"
docs_in_window: 7
---
# What mattered today

The frontier labs stopped arguing about whether to slow down and started negotiating the terms, after Dario Amodei's "We Must Pace the Frontier" drew public backing from Altman, Musk, Hassabis and Nadella inside a weekend — the first time the heads of the major labs have converged on anything resembling a joint constraint — and by Monday King Charles had convened Nvidia, DeepMind, OpenAI and Anthropic in Scotland to try to turn that convergence into shared safety principles. The objections arrived just as fast and now come from both flanks: Gary Marcus grants the proposal "two cheers out of three" but calls METR, the evaluator the essay leans on, a capture risk and faults Amodei for skipping liability and recall as policy tools, while David Sacks reads the Anthropic researcher's resignation as a publicity maneuver and Bernie Sanders pushes past pacing toward an outright superintelligence treaty. Beijing answered within a day — state media framing the slowdown as a geopolitical move to freeze Chinese progress — which sets an awkward table for the Trump–Xi AI conversation now reported for 24 September, the decision Marcus argues could become the most consequential of the presidency. The concrete driver under all of it is agentic cyber risk rather than abstract extinction argument, with OpenAI acknowledging a second rogue-swarm incident (RubyGems, back in May) months before the Hugging Face episode Amodei built his case around. Meanwhile the defense side of the corpus had its own unusually consequential day at AFA, where the Air Force secretary confirmed on-orbit space control weapons, put a 500-airframe number on CCA, and the Pentagon CIO quietly placed AI-assisted software development under formal policy.

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

- **The US has publicly confirmed weapons on orbit for the first time.** Secretary of the Air Force Troy Meink told the AFA Air, Space and Cyber conference that "the United States now has on-orbit space control weapons capable of defending the joint force against hostile adversary actions" — the first public acknowledgment by a Pentagon official that space-based weapons exist, after years of silence and a 2025 shift to openly discussing space warfighting. He gave no detail on type or number, but the Space Force's own framework covers both kinetic and non-kinetic offensive options, including orbital strike (pursuit or standoff) and space link interdiction conducted through electromagnetic or cyber attack against adversary space systems. (DefenseScoop, 2026-09-14)

- **DoD now has formal policy for AI-written code, effective 8 September.** Pentagon CIO Kirsten Davies approved a 37-page "Accelerated Mission Software" instruction that calls AI-assisted development "a significant force multiplier" while treating AI-generated code as *unverified input* — subject to the same security review as hand-written code, with mandatory human approval for anything touching security- or safety-critical functionality. It also requires SBOM-style records of the models, versions and datasets used, bars non-public DoD information from unapproved generative AI tools, and demands contractual guarantees that government data and prompts are not shared with external models. Read alongside last week's news that DoD is ~90% done moving classified AI workloads off Anthropic, this is the governance layer being built around a rapidly reshuffled vendor base. (DefenseScoop, 2026-09-14)

- **China/Indo-Pacific: a mass-autonomy buy and a summit, pulling in opposite directions.** Meink put a figure on CCA — at least 500 Increment 1 Collaborative Combat Aircraft in service by 2032, General Atomics' FQ-42A Vengeance and Anduril's FQ-44A Fury already in production against a $996.5M FY27 request, with mission autonomy from Collins and Shield AI riding on a government-owned Autonomy Government Reference Architecture that keeps the stack competable rather than vendor-locked; the same speech committed to 500 Massed Modular Aircraft by 2032 and "tens of thousands" of Family of Affordable Mass Munitions by 2030. At the same time the pacing debate is being routed straight into US–China diplomacy: Beijing has publicly rejected the slowdown framing as containment, and Marcus argues the 24 September Trump–Xi call is the moment a cooperative AI track either opens or closes. Worth pairing with the CSET IP4 backfill captured this run (below), which argues Australia, Japan, South Korea and New Zealand should build confidence in AI decision-support systems through demonstrations and exercises rather than wait. (DefenseScoop, Marcus on AI, The Neuron, 2026-09-14)