Algorithmic Diplomacy: A Policy Framework for AI-Mediated Interstate Negotiation

Policy Paper ● AI-mediated diplomacy is a human-controlled process where states encode constraints and preferences into computational systems, AI explores feasible agreement spaces, and human representatives authorize all outcomes—the framework augments rather than replaces diplomatic judgment. ● Pilot deployments should prioritize measurable domains (trade negotiations, climate agreements) where preferences are quantifiable and institutional frameworks exist, while avoiding core security bargains until verification and trust infrastructure matures. ● Baseline safeguards are non-negotiable: complete audit logs of constraint applications, explicit human authorization gates before any proposal is communicated, and clear responsibility assignment for every AI-generated option. ● Interoperability standards for "diplomatic AI" systems—covering minimum disclosure requirements, protocol rules, and mandatory red-team testing—are essential to prevent asymmetric advantage and manipulation between states with varying technological capabilities. ● A 12-month structured pilot program with defined metrics (negotiation time, option quality, perceived legitimacy) should be established through an existing multilateral framework, with results published to enable evidence-based scaling decisions.