← All signals
Governance

Synthetic Diplomacy

States war-game with machines, and the quiet work of negotiation begins to happen, in part, between their models.

What is it?

Governments feed AI into the machinery of statecraft — simulating adversaries, drafting positions, war-gaming outcomes before officials commit to them. The tooling augments analysts today; the pitch is speed nobody wants to be first to skip. Whether it stays advisory, or starts deciding, is the real fork.

Why now?

Statecraft always ran on scarce things — time to model an adversary, staff to game every branch of a negotiation. Cheap, abundant machine reasoning removes that ceiling, so more scenarios get run, faster, by fewer people. The capability is arriving well ahead of any doctrine for using it.

Who’s building it?

Defence and foreign-policy bodies commission the simulations; specialist AI vendors and analysts build and run them.

Defence ministriesForeign-policy bodiesSimulation vendorsNational labsOversight bodies
The full read
Why now, who is building it, where the market goes, and what stays unsettled.
Get weekly →
01 · The market
Who’s in it
Builders — specialist vendors and the defence primes behind them.
Buyers — defence and foreign ministries — few in number, each contract high-stakes.
Gatekeepers — security clearance regimes, procurement offices, and legislative oversight bodies.
What’s at stake
The speed that helps analysts narrows room for human deliberation — oversight and pace pull against each other.
Reputational exposure is high when a model misreads intent.
Vendor access to classified reasoning is a serious liability.
What’s unsettled
Whether this augments human judgment, or quietly starts to replace it.
Whether frontier labs or defence-specialist primes capture the value.
Whether model-shaped analysis measurably changes real decisions, or stays confined to exercises and drafts.
02 · How the demand matureda progression, not a timeline
Pilot
Analysts trial AI-drafted scenarios inside existing wargaming and simulation exercises.
Workflow
The tooling becomes routine — drafting positions and pressure-testing them before officials commit.
Oversight
Legislatures and inspectors general start asking who is accountable when a model-shaped brief informs an escalation call.
Market
Enough recurring contracts exist to justify dedicated defence-AI vendors, audit trails, and formal doctrine.
03 · Where the market goesthe forks, not a prediction
→Advisory tool or decision authority — decided by how far officials let model output substitute for judgment when time runs short.
→Frontier commercial labs may win the contracts, or classification barriers may keep this a specialist market.
→Allied governments may converge on shared simulation standards, or each build its own closed system.
→Escalation risk stays confined to exercises, or starts shaping decisions no one signed off on.
The open question is whether machine-modeled judgment stays advisory — or quietly becomes the thing officials lean on when a decision must move fast.
04 · The Agentic Delegation deskadjacent markets, moving now
Generative Governance
State agencies wire AI into workflows.
Pennsylvania expanded generative AI access to 3,000+ employees.
StateScoop
Machine Mediation
AI now mediates disputes once handled by courts.
CaseCraft AI launched a UK small-claims platform.
Legal IT Insider
Regulated Reality
Governments now legislate directly against synthetic deception.
30 US states now have election-deepfake laws.
Public Citizen
05 · The Wire
01
AI wargame teams identified 40% more risks than analog (2025).
Small Wars Journal ↗
02
AI chatbots chose nuclear escalation in 95% of war games (2026).
Euronews ↗
03
Code Metal lands $80 million Department of War contract to AI-enable wargaming
Fortune ↗
The panel on this signal
They agreed on
–Adoption today augments existing wargaming and analysis workflows inside government, not autonomous machine-to-machine negotiation between states.
–Value concentrates with vendors who clear defence procurement.
–Accountability for a model-informed, high-stakes call is the real unresolved seam.
The regulatory analyst disagreed on
–Documented use to date consists of simulation exercises and analytic support tools feeding human negotiators; no recorded instance shows a model exercising autonomous decision authority in an actual negotiation.
Know more about the method →
✦How this signal was built
01 · Panel
A Delphi panel of 4 AI analysts, each reasoning independently.
→
02 · Debate
They challenge and revise each other’s reads over two rounds.
→
03 · Evidence
Every claim is tied to a cited, linked source on The Wire.
→
04 · Synthesis
A moderator writes it into one read, disagreement and all.