Finkel Proposes AI Ethics Framework Amid Fragmented Global Governance

The proposal of modern "Three Laws for AI" by Alan Finkel arrives as global governments grapple with a fractured and reactive approach to AI governance. While the EU AI Act and recent US restrictions on frontier models represent initial steps, they remain far from the cohesive, principle-based guardrails Finkel envisions. This call for Asimov-style embedded ethics fundamentally reframes the AI safety debate, moving it from voluntary corporate pledges and post-hoc regulation toward a discussion about core architectural constraints, directly challenging the "move fast and break things" ethos that has powered the generative AI boom. Implementing such "laws" would fundamentally alter the AI value chain, creating a deep divide between auditable, "aligned" AI systems and the burgeoning open-source ecosystem. The winners would be large, vertically-integrated tech giants like Google and Microsoft, who can control the entire stack from silicon to software, making it easier to implement and verify such constraints. The losers are the disruptors—open-source leaders like Meta and Mistral AI, and hardware-agnostic model developers—whose competitive advantage lies in rapid, unrestricted iteration. This forces a strategic recalculation for frontier labs, shifting the focus from pure capability to provable safety. The debate Finkel’s proposal ignites will shape policy discussions in Washington and Brussels for the next 12 months, but the direct implementation of embedded "laws" remains a distant prospect. The more immediate consequence will be a market bifurcation: highly-regulated, expensive "certified" AIs for critical sectors like finance and defense, and a vast, innovative, but unregulated open-source landscape. The critical variable is whether a major AI-driven incident occurs, forcing governments to transform this philosophical exercise into a national security mandate. The real test is not if we can write the laws, but if we have the political will and technical capacity to enforce them.