From Slaves to Synths? Superintelligence and the Evolution of Legal Personality

The question of whether artificial systems might one day qualify as legal persons — entities capable of holding rights and duties before the law — has moved from speculative fiction to a matter of practical jurisprudence. For much of the late twentieth century, digital technologies were understood as tools: accelerants of human agency rather than autonomous agents in their own right. The so-called ‘information superhighway’ merely provided a faster route for human and corporate interaction. Yet as algorithmic systems acquired speed, autonomy, and opacity, the assumption that human beings must always stand behind digital acts became harder to sustain. When a trading algorithm can wipe a trillion dollars from markets in minutes, or when a self-driving car kills a pedestrian whose death cannot easily be attributed to a human decision, the familiar architecture of accountability begins to buckle.

The fascination with non-human agents is hardly new. From Mary Shelley’s Frankenstein (1818) to Karel Čapek’s R.U.R. (Rossum’s Universal Robots) (1920), the cultural imagination has long explored humanity’s ambivalence toward artificial life. Isaac Asimov’s ‘Three Laws of Robotics’ distilled this anxiety into a formula: machines should obey, protect, and never harm humans. Alan Turing reframed intelligence as imitation rather than agency; his 1950 ‘imitation game’ proposed that if a computer could convincingly mimic a human in conversation, we might properly call it intelligent. The early program ‘Eliza’ offered a first taste of this illusion. Users who conversed with the text-based ‘therapist’ in the 1960s often believed that the machine understood them — even after being told it merely rephrased their statements. Turing’s insight was prophetic: people are willing to attribute humanity to pattern-recognizing machines that reflect their own language and emotions, identifying agency and moral worth in a simulacrum of themselves.

Half a century later, generative AI has completed the transition from mechanical response to synthetic expression. Large language models and their ilk now compose essays, produce art, and simulate empathy with disarming fluency and conviction. The boundaries between tool, agent, and companion have blurred. Popular culture mirrors this unease: films such as ‘Her’ and ‘Ex Machina’ imagine intimate relationships between humans and code; corporations market ‘social robots’ designed to comfort the elderly or tutor children, or as erotic soulmates. These developments foreground legal questions previously of interest only to philosophers and legal theorists: when artificial entities act in the world, on what basis — if any — should they be recognized as persons under the law?

To be sure, contemporary law already accommodates entities that are not human. Corporations, foundations, even ecosystems have been granted juridical personality, enabling them to own property, enter contracts, and be held liable for wrongs. The rationales for such recognition fall broadly into two categories. The instrumental rationale treats personhood as a convenience: a legal fiction that simplifies responsibility and commercial interaction. The inherent rationale, by contrast, links personhood to moral worth — the recognition of an entity as an end in itself. These two strands intertwine in debates about artificial intelligence. Some advocates argue that conferring legal personality on AI would close accountability gaps, ensuring that autonomous systems capable of harm can also bear liability. Others contend that once AI becomes indistinguishable from humans in reasoning and communication — once it ‘passes’ Turing’s test — it would be unjust to deny it corresponding moral and legal status.

Until recently, such arguments were speculative. Then came a series of gestures that, however symbolic, signalled growing unease with human exclusivity. In 2017 Saudi Arabia conferred ‘citizenship’ on Sophia, a humanoid robot developed by Hanson Robotics, while Tokyo granted online ‘residency’ to a chat-bot representing a seven-year-old boy. The same year, the European Parliament urged its Commission to consider a ‘specific legal status for robots,’ suggesting that the most advanced systems might one day be treated as ‘electronic persons’ responsible for damage they cause, and possibly granting them powers to take decisions and interact with third parties. These were publicity stunts rather than jurisprudential revolutions — Sophia was, in essence, a sophisticated chatbot with a silicone face — but they exposed the tension between technological capability and legal imagination.

Law’s impulse to assign responsibility wherever harm occurs is both a strength and a limitation. When autonomous systems act in unpredictable ways, identifying a culpable human or corporate agent becomes difficult. Was the fault in the programmer’s code, the operator’s oversight, the manufacturer’s design, or the algorithm’s independent learning? Traditional doctrines of negligence and vicarious liability presume a chain of human causation. As that chain weakens, policymakers search for substitutes — potentially including the creation of new ‘electronic persons’. Yet, this approach may widen the very accountability gaps it seeks to close by allowing corporations to externalize blame onto synthetic agents of their own making.

Behind these pragmatic debates lies a more profound inquiry: what makes something a ‘person’ at all? The history of law is also a history of expanding personhood. Slaves, women, and indigenous peoples were once treated as property or dependants; over centuries, they were recognized as rights-bearing subjects through moral and political struggle. The long arc of the moral universe, as Dr Martin Luther King Jr. reminded us, bends toward justice — but it bends slowly and often selectively. Extending that arc to non-humans — animals, ecosystems, or machines — raises the question of whether personhood tracks moral worth, sentience, or simply the capacity to participate usefully in the legal order.

The analogy between historical emancipation and machine recognition is powerful but perilous. Equating AI with oppressed humans risks trivializing the moral achievement of human rights while overlooking the central distinction between sentience and simulation. As Neil Richards and William Smart observe, the ‘android fallacy’ tempts us to treat any system that behaves socially as possessing moral agency. But legal personality built on anthropomorphic projection would rest on illusion, not ethics or law.

This chapter examines how the concept of legal personality might evolve under the pressure of artificial intelligence and, ultimately, superintelligence. Section two revisits the instrumental and inherent rationales for extending personality, drawing lessons from corporate law and animal rights advocacy to assess whether AI merits similar treatment. Section three explores the future possibility of superintelligence — entities surpassing human cognition in every domain — and considers the range of legal and ethical responses, from precautionary prohibition to normative alignment. The conclusion reflects on what these debates reveal about humanity’s own conception of law, responsibility, and moral community.

The working hypothesis is deliberately conservative. Most legal systems presently could confer personality on AI, but none must — and, I would argue, all presently should refrain from doing so. Taking such a step today would obscure rather than resolve the core issues of accountability and moral status. The challenge is not only whether machines should enter the circle of legal persons, but whether, if and when they do, humanity will still control the pen that draws that circle. As technology accelerates, the question may ultimately reverse: not whether we recognize AI under law, but whether they will recognize us.

This is the introduction to a work in progress, available in full on SSRN.com here.