US AI leaders seek regulation, but the Trump administration remains dismissive. The UK’s AI Security Institute (AISI) emerges as a superior solution for evaluating AI risks. Well-funded with expert staff and a track record of testing advanced models, AISI offers independence from Silicon Valley's commercial pressures. The author advocates for AISI to become the primary, government-backed evaluator for AI, arguing it should be granted enforcement powers. This would establish a clear path from testing to regulation, mirroring oversight in other high-risk sectors like atomic energy and aviation, ensuring critical safety as AI progresses.