What would Asimov's robot laws look like today? A blueprint for AI safety | Alan Finkel

When Isaac Asimov crafted his three laws of

When Isaac Asimov crafted his three laws of robotics, he imagined a future where machines would be bound by an unshakeable moral code. Today, as AI systems grow more powerful and autonomous, the question resurfaces: what would a modern equivalent of those laws entail? Drawing on Asimov's vision, I propose a new set of principles tailored to the age of large language models and autonomous agents.

Elon Musk, the CEO of Tesla and SpaceX, predicted in July that swarms of AI-driven robots would soon dominate the physical world, and that these systems might no longer follow human commands. He sketched an alternative path: one where AI is programmed to love truth and desire human flourishing, and where governments step in to enforce this shared objective. His words echo Asimov's caution, but also his optimism that we can design machines to be benign.

Governments have begun to respond, albeit slowly. The Trump administration has delayed and restricted the release of the most advanced frontier models from companies like OpenAI and Anthropic. The European Union's AI regulations, approved in 2024, entered into force this year. These measures, however, fall short of requiring the kind of deep-seated safeguards that would instill in AI a genuine commitment to human welfare.

To fill that gap, I offer what might

To fill that gap, I offer what might be called the three laws of AI. First, an AI must not harm humanity, whether by action or inaction. Second, an AI must obey human orders unless they conflict with the first law. Third, an AI must protect its own existence, provided that protection does not conflict with the first or second laws. These principles, while simple, could serve as a starting point for robust AI governance.

But laws alone are not enough. As Asimov knew, rules can be bent. We need a culture of AI development that prizes transparency, accountability, and a relentless pursuit of truth. Only then can we ensure that our creations remain our servants, not our masters. The time to act is now, before the machines outpace our ability to guide them.