US–China Superintelligence Bilateral Protocol
Explore actionable treaty architectures between Washington and Beijing. Calibrate frontier compute ceilings, cryptographic hardware attestation, autonomous strike command bans, and joint incident verification to measure strategic crisis stability.
Bilateral Stability Assessment
Evaluated against game-theoretic breakout modelsOperative Articles of Accord (Summary Draft)
Strategic Mechanics of Superintelligence Accords
When great powers meet to negotiate on frontier artificial intelligence and "super intelligence," traditional arms control treaties (such as START or the CWC) fall short because algorithmic models can be copied in seconds and weights transferred invisibly across fiber-optic cables.
Consequently, realistic diplomatic frameworks pivot on physical compute chokepoints: the gigawatt electrical footprints of hyperscale clusters, semiconductor supply chains, and on-die cryptographic co-processors that can prove training run limits without exposing proprietary national model weights.
Frequently Asked Questions
What does "Hardware Root-of-Trust Attestation" mean?
Leading microchips (GPUs/TPUs) include physically unclonable functions and secure enclaves. Under a verifiable treaty, chips can cryptographically sign telemetry logs proving they have not participated in training runs exceeding agreed-upon floating-point operation thresholds (FLOPs).
Why is Breakout Time the central metric?
In nuclear strategy, breakout time is the duration required to enrich weapons-grade fissile material. In superintelligence strategy, breakout time is the months required for an adversary to secretly build a clandestine training cluster and train a transformative frontier system before detection and diplomatic response.
Can civilian economic AI be separated from frontier national security runs?
Yes. The majority of commercial, medical, and consumer AI operates far below 10²⁵ FLOP. By setting high thresholds for treaty-level reporting, domestic enterprise innovation proceeds unfettered while strategic foundation model training clusters remain transparent.