What happened
RAND published a new economic framework modeling how the cost of reproducing a demonstrated frontier AI capability falls over time as hardware improves, algorithms become more efficient, and knowledge diffuses. Authored by Tobias Sytsma, the report formalizes what this cost-decay dynamic means for governance regimes attempting to keep dangerous AI capabilities away from untrustworthy actors, and quantifies the shrinking 'window of opportunity' policymakers have to act on counterproliferation before capabilities become cheap and widely accessible. The framework offers policy implications for export controls, model security requirements, and the timing of nonproliferation interventions relative to capability diffusion curves.
Why it matters
Gives national-security and export-control policymakers a structured economic lens for timing interventions (chip controls, model-weight security mandates) before frontier capabilities commoditize — directly relevant to ongoing US-China AI competition debates.
Action needed
Brief national-security and trade-policy teams on the counterproliferation window framework to inform timing of export-control and model-security advocacy.