For the past thirty years, governments have grappled with the challenge of controlling the export of cybersecurity-related software. The underlying assumption has been that by limiting access, they could curb the proliferation of potentially harmful technologies. However, history suggests this approach has been largely ineffective. Despite numerous regulations and restrictions, the flow of such software has continued unabated, often finding its way to intended recipients through various channels. This long-standing pattern casts doubt on the viability of current strategies, especially in the context of rapidly evolving technologies like advanced AI models. The recent development of Anthropic's cybersecurity model, Mythos, prompts a critical re-evaluation: if export controls have failed to stem the tide of traditional cybersecurity tools, why should we expect them to succeed with sophisticated AI-driven solutions? The focus may need to shift from restrictive export policies to proactive measures that foster responsible development and international collaboration.