CEO Dario Amodei cautioned Congress that releasing powerful AI models without centralized oversight creates irreversible risks. Western enterprises are now adopting open-weight Chinese models as they close the performance gap with US systems. This shift bypasses domestic safety guardrails. Practitioners must now weigh raw performance against the lack of centralized governance in these models.