Understanding the threat. Countering with sovereign capability. Open intelligence analysis of China's military AI ambitions and how the UK's sovereign defence OS provides an asymmetric counter.
This is an unclassified analytical page based on publicly available sources: PLA writings, China defence white papers, open-source intelligence, US DoD reports to Congress, and IISS/RUSI analyses. Nothing here is classified. Everything here is the open conversation we need to have about UK sovereign AI defence.
China's People's Liberation Army has the most explicit and well-funded military AI programme in the world. The 2017 Party Congress elevated "intelligentised warfare" (ζΊθ½εζδΊ) to the level of doctrine β making China the first major power to formally integrate AI into military operations at the strategic level.
| Capability | China (PLA Assessment) | DEFONEOS Counter | UK Advantage |
|---|---|---|---|
| Intelligentised Warfare | AI integrated into all PLA operations; made doctrine at 19th Party Congress 2017 | 12-domain OS covers all UK defence domains | Open source = allied verification. China's system is opaque; ours is transparent. |
| AI Weapons Systems | Autonomous drones, AI-guided missiles, swarm boats (Yantai, 2020: 56-vessel swarm) | Counter-drone MCP + Batear acoustic detection + RF jamming + swarm simulation (Mava) | Defensive focus. We don't need to out-swarm China β we need to counter their swarm. |
| DeepSeek / Huawei / Baidu Models | Sovereign AI models (open weights, government-backed). DeepSeek-V3, PanGu, ERNIE | SOV3 + qwen3:30b-a3b (sovereign, local, offline-capable) | We run sovereign models on UK hardware. No CLOUD Act exposure. |
| PLA Strategic Support Force (SSF) | Cyber, space, electronic warfare integrated command (est. 2015) | DEFONEOS integrates same domains + 6 more (cognitive, civil, compliance, etc.) | BFT governance. China has no equivalent of a 33-agent council voting on operations. |
| Cognitive Domain Operations (CDO) | PSYOPS via social media + AI-generated content + narrative warfare. Active in Taiwan, Pacific, Africa. | Cognitive domain (7 systems) + GDELT MCP + narrative detection + AI-generated content detection | Democracies are more resilient to CDO when citizens are informed. DEFONEOS provides the transparency. |
| AI Standards Setting | Setting global AI governance standards via ITU, ISO. "China Standards 2035" project. | CSOAI compliance framework + JSP 936 + EU AI Act alignment + open charter | UK as the democratic alternative standard-setter. Open source > closed control. |
| Quantum | Quantum communication network (Beijing-Shanghai), quantum radar claims | PQC migration roadmap (ML-KEM-768 / ML-DSA-65), quantum-safe SIGIL chain | NIST PQC standards adopted. Quantum-safe by design. |
| Hypersonic | DF-17 hypersonic glide vehicle, DF-ZF | Sensor detection (infrared satellite + radar), early warning, BFT decision support | DEFONEOS doesn't build weapons. It builds the decision-support that lets UK defences respond. |
The threat: UK defence currently depends on US cloud infrastructure (AWS GovCloud, Azure Government). Under the US CLOUD Act (2018), US providers can be compelled to disclose data stored on their infrastructure, even if hosted in the UK. China's intelligence services actively target US cloud supply chains.
The DEFONEOS counter: Full sovereign deployment β all data, all compute, all models on UK soil. JSP 440 compliant. No CLOUD Act exposure. Hardware options include sovereign data centres, air-gapped M2/M4 edge nodes, or sovereign GCP regions (London).
| Deployment Mode | Infrastructure | CLOUD Act Exposure | Use Case |
|---|---|---|---|
| Air-gapped (M2/M4) | Apple Silicon edge nodes, local SSD | ZERO | Deployed operations, field HQ |
| Sovereign cloud (UK) | UK-only GCP/AWS region, encrypted at rest | NONE (sovereign tenancy) | Command centre, back office |
| On-premise data centre | MODNET, sovereign DC, HPE/Dell | ZERO | Fixed installation, strategic HQ |
| US cloud (non-defence use) | AWS/Azure standard | EXPOSED | Development only, no operational data |
The threat: China's advantage is scale β 1.4 billion people, state-directed AI investment, massive data collection without privacy constraints. The UK cannot match China on scale. But we can win on trust.
The DEFONEOS counter: Open source under Apache 2.0. Allied nations can read every line of code, verify every algorithm, and audit every decision. China's systems are opaque; ours are transparent. When Five Eyes, AUKUS, and NATO allies deploy DEFONEOS, they know exactly what it does β because they can read the code.
| Dimension | China PLA AI | DEFONEOS |
|---|---|---|
| Source code | CLOSED State secret | OPEN Apache 2.0, auditable |
| Algorithm transparency | OPAQUE | TRANSPARENT All models documented |
| Decision audit trail | NONE externally verifiable | SIGIL Ed25519 + BTC anchored |
| Independent verification | IMPOSSIBLE | DESIGNED FOR IT defoneos-sigil verify |
| Allied trust | Low β "black box" | High β "glass box" |
The threat: China's Three Warfares doctrine includes psychological warfare β targeting Western populations via social media manipulation, AI-generated content (deepfakes, bot armies), and coordinated narrative operations. Active in Taiwan elections, Pacific island nations, and increasingly in Europe.
The DEFONEOS counter: The cognitive domain MCP suite detects and counters information operations in real-time:
| Cognitive Defence MCP | Function | How It Counters CDO |
|---|---|---|
gdelt-news-mcp | Global news event monitoring (GDELT Project) | Detects coordinated narrative pushes across 65K+ news sources |
social-media-mcp | Social media analysis (API-gated) | Identifies bot networks, coordinated inauthentic behaviour |
cdo-detect-mcp | AI-generated content detection | Deepfake detection, synthetic text classification |
narrative-tracker-mcp | Narrative evolution tracking | Maps how disinformation narratives spread and mutate |
influence-scoring-mcp | Influence operation scoring | Assesses the impact and origin of influence campaigns |
Key insight: Democracies are MORE resilient to cognitive warfare when citizens are informed about the operations targeting them. DEFONEOS's open approach means the public can see what's being detected β transparency is the antidote to manipulation.
The threat: China leads the world in military swarm technology. The October 2020 Yantai demonstration deployed a 56-vessel autonomous swarm. The PLA has tested drone swarms with 100+ units. These overwhelm traditional point-defence systems.
The DEFONEOS counter: Layered, multi-modal counter-swarm capability:
| Layer | System | Counter-Swarm Capability |
|---|---|---|
| Detection | Batear acoustic + RF + radar + EO/IR | Detect swarm at 5-15km depending on altitude |
| Classification | YOLOv8 + DeepStream + model fusion | Classify drone type, assess payload, determine intent |
| Soft kill | RF jamming (GPS + comms link) | Disrupt swarm coordination. Force RTL (return to launch). |
| Hard kill | Interceptor drones (autonomous) | Physical interception of threat drones |
| Swarm simulation | Mava + PX4 + SITL | Model adversary swarm tactics. Train counter-swarm agents. |
| Decision support | BFT council + SOV3 intuition | Evaluate counter-swarm options. Proportionality check. |
Asymmetric advantage: We don't need to out-build China's swarm capability. We need to out-counter it. A defensive counter-swarm system is cheaper, faster to develop, and more aligned with UK defence posture than an offensive swarm arms race.
The threat: China is actively setting global AI governance standards through ITU, ISO/IEC JTC1/SC42, and the "China Standards 2035" project. If China sets the standards for military AI, democratic values lose.
The DEFONEOS counter: Position the UK as the democratic alternative standards leader. DEFONEOS provides a working, open, deployed system that allies can adopt β with governance, audit trails, and ethical constraints built in. Standards set by working code are more powerful than standards set by committee.
| Standard Area | China's Push | DEFONEOS / UK Position |
|---|---|---|
| AI audit trails | State-controlled, not externally verifiable | SIGIL chain β open, Ed25519-signed, BTC-anchored, court-admissible |
| AI governance | Party oversight, state control | BFT 33-agent council, Charter-governed, democratic values |
| AI transparency | Opaque | Apache 2.0 open source β the most transparent defence AI in the world |
| Lethal autonomous systems | Ambiguous β developing but not committing to LAWS ban | Charter Article 7: NEVER. Clear red line. Democratically accountable. |
| Data sovereignty | State access to all data (National Intelligence Law 2017) | UK sovereign infrastructure, DPA 2018 / GDPR compliant, minimal data |
| Capability | China PLA | UK with DEFONEOS | US (Palantir/Anduril) |
|---|---|---|---|
| Military AI doctrine | YES Intelligentised warfare | YES 12-domain sovereign OS | FRAGMENTED Vendor-led |
| Sovereign infrastructure | YES All domestic | YES UK-only, JSP 440 | US CLOUD Act |
| Open source | NO State secret | YES Apache 2.0 | NO Proprietary |
| Audit trail | OPAQUE | SIGIL BTC-anchored | Limited Internal logs only |
| Governance | PARTY CCP oversight | BFT 33-agent council | CORPORATE Board oversight |
| LAWS prohibition | AMBIGUOUS | ABSOLUTE Charter Art 7 | POLICY Not code-enforced |
| Allied interoperability | None (non-aligned) | AUKUS + Five Eyes + NATO | YES But US-controlled |
| Scale | MASSIVE 1.4B population | Lean (startup) | LARGE $20B+ revenue |
China's AI advantage is scale and centralisation. DEFONEOS's advantage is trust and transparency. In a world where allies must cooperate, trust beats scale. When Five Eyes and AUKUS partners deploy a system they can audit, verify, and govern collectively, the democratic alliance has something China can never have: mutual trust in a shared defence infrastructure.
| Source | Topic | Reference |
|---|---|---|
| US DoD Annual Report to Congress | China military power, AI capabilities | "Military and Security Developments Involving the PRC" (annual) |
| IISS Military Balance | PLA order of battle, modernisation | The Military Balance 2025 |
| RUSI | China's cognitive warfare, AI standards | RUSI Occasional Papers (various) |
| CSET Georgetown | China AI investment, military-civil fusion | Center for Security and Emerging Technology reports |
| NSCAI Final Report | US strategy for AI competition | National Security Commission on AI (2021) |
| China Defence White Paper | China's stated military doctrine | "China's National Defence in the New Era" (2019) |
| PLA Daily / China Military Online | PLA writings on intelligentised warfare | Open-source monitoring |