If the U.S. opts for a unilateral approach, it must replace international consensus with internal, high-friction barriers. One method involves implementing strict liability frameworks. By making developers legally responsible for downstream harms caused by autonomous agents, the law forces companies to internalize the costs of catastrophic failures. This shifts safety from a voluntary ethical guideline to a core financial necessity.
Another lever is the control of hardware bottlenecks. The U.S. can use export controls on advanced semiconductor manufacturing equipment and high-end GPUs to regulate who can build the most capable models. This creates a physical boundary on computational power, making it difficult for unsafe actors to scale risky architectures regardless of the lack of global rules.
Finally, the government could mandate rigorous, pre-deployment testing for models exceeding specific FLOP thresholds. Think of it like FAA certification for aviation. By requiring third-party red-teaming and hardware-level telemetry, the state ensures that a rush to market does not bypass fundamental stability checks. These domestic guardrails create a high bar for participation, ensuring that even if neighbors lower their standards, the American market demands safety to operate within its borders.
