2028: Two Scenarios for Global AI Leadership
Anthropic Policy, Thursday, May 14th, 2026
The US must strengthen AI export controls to maintain democratic leadership over authoritarian regimes by 2028.
Anthropic outlines two scenarios for US-China AI competition by 2028, emphasizing that democracies must lead in AI development to prevent authoritarian regimes from using powerful AI systems for repression and military advantage.
The critical factor is compute access-the computer chips needed for training advanced AI models, which are currently supplied by American companies. China has kept pace by exploiting loopholes in export controls and conducting large-scale distillation attacks to extract innovations from American models.
If policymakers tighten export controls, disrupt distillation attacks, and accelerate democratic adoption of AI, democracies can establish a commanding 12-24 month lead by 2028. Without action, the CCP could reach competitive capabilities, allowing authoritarian regimes to set global AI norms and enable automated repression at unprecedented scale.