Authored By: Okusanya Esther Oluwatomilola
University of Lagos
Introduction
In January 2024, singer Taylor Swift became the target of sexually explicit AI-generated deepfakes that spread rapidly across social media platforms.1 The incident sparked worldwide concern about the harmful consequences of artificial intelligence and renewed debates on regulation and accountability. If artificial intelligence can create harmful content capable of damaging an individual’s privacy and reputation, a more difficult question arises: when autonomous artificial intelligence systems make decisions that cause harm, who should bear the responsibility — the programmer, the manufacturer, the user, or the artificial intelligence system itself? This question forms the basis of the debate on AI personhood and the attribution of responsibility.
Artificial intelligence has transformed modern society into a technologically advanced one and has emerged as a transformative force across various sectors, reshaping economic landscapes, social interactions, and ethical considerations. Advanced AI systems are capable of learning, analysing information, and making decisions with minimal human intervention. Although these developments have enhanced productivity and societal advancement, they have also created challenges regarding legal responsibility and accountability. Existing legal frameworks were primarily designed to regulate natural persons and artificial persons such as corporations, rather than autonomous technological systems capable of independent decision-making.
This article examines whether autonomous artificial intelligence systems should be recognised as legal persons capable of bearing responsibility for harmful outcomes resulting from their decisions. It argues that although AI systems increasingly perform functions resembling independent agency, responsibility should remain with the human actors involved in the design, development, deployment, and supervision of such systems, because AI presently lacks consciousness, moral understanding, and genuine legal agency. The article focuses primarily on legal and moral responsibility in relation to autonomous AI systems. It does not attempt to resolve broader philosophical questions concerning machine consciousness or the future development of sentient artificial intelligence.
The article is organised into five parts. First, it examines the legal framework surrounding artificial intelligence and personhood. Secondly, it analyses the concept of the responsibility gap created by autonomous AI systems. Thirdly, it evaluates arguments both supporting and opposing AI legal personhood. Fourthly, it considers comparative international perspectives on AI regulation. Finally, it proposes that legal systems should strengthen human-centred accountability frameworks rather than recognise AI as independent legal persons.
Artificial Intelligence, Personhood and the Responsibility Gap
Meaning of Artificial Intelligence
Artificial intelligence (AI) is technology that enables computers and machines to simulate human learning, comprehension, problem-solving, decision-making, creativity, and autonomy.2 AI systems are capable of performing some tasks that typically require human intelligence, such as understanding language, recognising patterns, and making decisions. AI influences various fields, including education, the social sciences, and information technology, and plays a significant role in promoting gender equality and advancing knowledge in the natural sciences. However, the expansion of autonomous AI systems has also created significant concerns regarding accountability where harmful outcomes occur without direct human intervention.
Legal and Moral Personhood
A proper analysis of liability in autonomous AI systems requires an understanding of legal and moral personhood. Legal personhood refers to the recognition by law of an entity capable of bearing rights and duties, such as owning property, entering into contracts, or suing and being sued. Traditionally, legal personhood extends beyond natural persons to include corporations and organisations recognised as artificial legal entities, allowing such entities to participate in the legal system in the same manner as natural persons.3 Moral personhood, on the other hand, is generally attributed to natural persons and entities possessing characteristics such as sentience, consciousness, self-awareness, and the ability to experience harm. These characteristics become significant in discussions regarding whether AI systems may qualify as persons capable of bearing legal responsibility.
Although AI systems may imitate human reasoning, they presently lack subjective experience and consciousness. Nonetheless, some scholars and policymakers argue that granting limited legal standing to AI, particularly in high-risk or highly autonomous contexts, could enhance transparency, facilitate liability attribution, and deter negligent design or deployment.4 Others caution that such recognition risks creating a legal fiction capable of shielding human actors from accountability, in much the same way critics argue corporate personhood has occasionally been misused.5
The Responsibility Gap
Autonomous AI systems also create what scholars describe as a “responsibility gap.” This gap arises where harmful outcomes become difficult to attribute to a specific actor because the system operates with a degree of independence from direct human supervision. The gap primarily exists where an autonomous system causes harm without any identifiable actor — the programmer, manufacturer, operator, or user — being directly responsible, and where this absence of responsibility results from the autonomous nature of the system itself. Commentators identify two conditions that must be satisfied for a responsibility gap to arise:
No Responsibility: An autonomous system carries out an action for which no one — neither the programmer, the manufacturer, the operator, nor the autonomous system itself — is responsible.
Autonomy: This absence of responsibility is due to the system’s autonomy.6
While some scholars argue that this gap justifies recognising AI as an independent legal entity, such an approach creates significant legal and ethical concerns. Responsibility within legal systems traditionally depends upon intentional conduct, moral understanding, and the capacity to appreciate consequences. AI systems presently operate through algorithms, training data, and computational processes rather than conscious reasoning. Consequently, attributing responsibility directly to AI risks weakening accountability by enabling corporations, developers, and operators to evade liability for harms caused by systems they created and deployed.
Arguments Opposing Granting AI Legal Personhood
Lack of Consciousness and Qualia
One of the strongest objections to AI personhood is that AI systems lack consciousness and subjective experience. Without qualia — the subjective “what it is like” aspect of experience — there can be no genuine consciousness and therefore no true moral standing.7 Under this view, even highly sophisticated AI systems remain philosophical zombies: systems capable of simulating intelligent behaviour without possessing internal awareness or subjective experience. AI systems neither feel nor suffer and therefore possess no intrinsic interests requiring legal protection. This distinction is significant because legal responsibility has traditionally been connected to intentional conduct and moral understanding rather than sophisticated automation.
Absence of Moral Agency
Another important objection concerns the absence of moral agency. Established laws and ethics are based on an individual’s understanding of consequences and moral choices. AI cannot truly be responsible for its actions because it operates through algorithms and programming and is designed, owned, and maintained by humans. Granting AI legal personhood could allow companies to shift responsibility to AI systems, leaving victims with limited avenues for recourse. Rather than strengthening accountability, AI personhood could weaken existing legal protections.
Human Dignity Concerns
Moreover, AI personhood may undermine human rights by diminishing the significance of unresolved human concerns and redirecting attention towards machines rather than human beings capable of suffering. AI systems are created for specific purposes, and granting rights to such systems could interfere with their intended functions while creating unnecessary restrictions on beneficial technologies. Consequently, the distinction between humans and machines may become blurred in a manner that threatens human dignity.8
AI as Pattern Recognition Rather Than Genuine Reasoning
Critics argue that AI does not truly reason but merely imitates reasoning through pattern recognition derived from training data, thereby rendering it incapable of agency or personhood. However, reasoning involves integrating multiple constraints, resolving conflicts, and applying knowledge in unfamiliar situations. Since humans similarly rely on learned experiences and patterns, the central issue becomes whether AI can apply such patterns coherently in new contexts.
Dependence on Human Control and Alignment
Finally, some argue that alignment reduces AI to a system designed merely to obey human instructions rather than function as an independent agent. Because AI is designed, trained, and maintained by humans, it remains dependent upon external control and operates within predetermined objectives and constraints. Some scholars argue that current laws are adequate to address AI-related issues, as manufacturers, suppliers, owners, and users may already be liable under tort law, product liability law, and principles of agency.9 Consequently, they argue that there is no need to establish a new category of AI personhood.
Arguments Supporting AI Personhood and Responsibility Attribution
AI as an Independent Decision-Maker
Despite these objections, some scholars and policymakers argue that increasing AI autonomy makes certain AI systems resemble independent decision-makers rather than ordinary technological tools. As AI systems increasingly generate outputs capable of influencing real-world outcomes, some scholars contend that responsibility should correspond to the operational role played by AI within decision-making processes.
Limited Legal Standing for AI Systems
Supporters of limited AI personhood argue that recognising certain autonomous systems as legal entities could improve transparency, facilitate liability attribution, and encourage responsible technological development.
Challenges to Traditional Human-Centred Accountability
Proponents further argue that advancing AI capabilities may eventually challenge existing human-centred models of accountability and expose the limitations of current legal categories. In situations where autonomous systems independently make decisions capable of causing significant harm, traditional liability frameworks may struggle to identify a single responsible actor. Some scholars therefore suggest that limited forms of AI legal recognition could operate similarly to corporate legal personality by allowing autonomous systems to bear certain legal obligations without necessarily granting them full human rights. However, these arguments remain controversial because AI systems presently lack the essential characteristics traditionally associated with legal and moral responsibility.
Global Perspectives on AI Personhood
The European Union has entertained the possibility of granting robots special legal status as “electronic persons” to ensure civil liability for damages, while maintaining an overall cautious stance. The increasing autonomy of robots renders ordinary liability rules insufficient, highlighting the need for new legislation. By contrast, the United States regards robots as property with no explicit legal rights, and the US Copyright Office denies copyright protection for works generated autonomously by AI without human authorship. In Asia, countries such as Japan and South Korea have explored ethical guidelines concerning AI, but stop short of considering its personhood, focusing instead on maintaining human control over AI development and use. There is no global consensus on this matter: the EU recognises the potential of legitimising a form of robot personhood, while the US, Japan, and South Korea remain focused solely on ethical governance. The discussion is ongoing and continues to be shaped by advances in AI technology and shifting societal attitudes.
Proposed Solutions and Recommendations
Strengthening Human-Centred Accountability
Rather than granting full legal personhood to AI systems, legal frameworks should strengthen human-centred accountability mechanisms capable of addressing harms caused by autonomous technologies.
Regulatory Oversight and Transparency
Regulatory frameworks should impose transparency and oversight obligations upon developers of high-risk AI technologies in order to preserve accountability and public trust. Such frameworks may include mandatory human oversight, auditing requirements, and stricter compliance obligations for high-risk AI systems.
Reforming Existing Liability Frameworks
Governments should develop clearer statutory rules governing liability for harms resulting from autonomous AI systems. Existing principles of tort law and product liability should also be modernised to accommodate the complexity of machine learning systems.
International Cooperation in AI Governance
International cooperation is equally necessary, because AI technologies increasingly operate across national boundaries. Unified global standards would help ensure accountability while reducing regulatory fragmentation and jurisdictional inconsistencies. These reforms would preserve meaningful human responsibility while allowing legal systems to adapt to emerging technological realities.
Conclusion
The debate surrounding AI personhood and responsibility remains unresolved due to the increasing autonomy of artificial intelligence systems and the limitations of existing legal frameworks. Although autonomous AI systems can independently influence real-world outcomes, they presently lack consciousness, moral understanding, and genuine legal agency. This article has demonstrated that existing concerns regarding accountability are better addressed through strengthened human-centred liability frameworks rather than by granting independent legal personhood to AI systems. While limited forms of legal recognition may eventually emerge for regulatory purposes, full AI personhood at the present stage risks creating responsibility gaps and enabling human actors to evade liability. Ultimately, the central challenge for modern legal systems is not whether machines should become persons, but how the law can preserve meaningful human accountability in an age increasingly shaped by autonomous technologies.
Reference(S):
1. ‘Pornographic Deepfakes of Taylor Swift Emerge on Social Media’ ABC News (26 January 2024) <https://www.abc.net.au/news/2024/01/27/how-ai-is-creating-taylor-swift-pornographic-deepfakes/103396284> accessed 20 May 2026.
2. IBM, ‘What Is Artificial Intelligence (AI)?’ (Cole Stryker and Eda Kavlakoğlu) [URL and access date needed — please confirm and insert].
3. ‘Legal Aspects of Artificial Intelligence Personhood’ (SSRN) <https://www.ssrn.com/abstract=4749785> [author and year needed — please confirm and insert].
4. R Dremliuga, P Kuznetcov and A Mamychev, ‘Criteria for Recognition of AI as a Legal Person’ (2019) 12(3) Journal of Politics and Law 105.
5. S Ripken, ‘Constitutional Dimensions of the Corporate Person: Corporate Free Speech’ in Corporate Personhood (CUP 2019) [full book title/editor needed — please confirm].
6. [Author], ‘Title’ (2022) 24 Ethics and Information Technology 36 <https://doi.org/10.1007/s10676-022-09643> [DOI appears incomplete and author/title are missing — please confirm and insert].
7. S Ripken (n 5).
8. A Birhane, J van Dijk and F Pasquale, ‘Debunking Robot Rights Metaphysically, Ethically, and Legally’ (2024) arXiv [arXiv identifier/URL needed — please confirm and insert].
9. G Wagner, ‘Robot, Inc: Personhood for Autonomous Systems?’ Fordham Law Review <https://ir.lawnet.fordham.edu/cgi/viewcontent.cgi?article=5632&context=flr> accessed 19 February 2025 [publication year needed — please confirm].
