From Code To Creation: Ownership Of AI Systems And AI-Generated Works

Introduction

From credit scoring tools and fraud detection systems used by financial institutions to recruitment software, health diagnostics, and content moderation systems, Artificial Intelligence (“AI”) is increasingly becoming integrated into everyday business operations in Nigeria. Businesses across multiple sectors are rapidly adopting AI-driven technologies; startups now deploy machine learning tools to personalise user experiences, while creative industries, regulators and the general public continue to grapple with the implications of AI-generated content and automation.

As AI technologies gain commercial relevance, two fundamental legal questions continue to emerge. First, who owns the code, models, datasets, and algorithmic components underlying AI systems? Second, who bears responsibility where an AI system produces harmful, discriminatory, or unlawful outcomes?

These issues do not fall neatly within a single area of Nigerian law. Rather, they exist at the intersection of intellectual property law, data protection, contract law, labour law, tort, consumer protection, and Nigeria’s developing regulatory approach to AI governance. In the absence of comprehensive AI-specific legislation, ownership, control, and accountability are largely determined through existing legal frameworks and contractual arrangements between parties.

Intellectual Property and Ownership of AI Systems

Under Nigerian law, the Copyright Act¹ states that computer programmes qualify as literary works and are capable of copyright protection, provided they are original and fixed in a tangible medium of expression.² However, copyright protection does not extend to ideas, mathematical concepts, methods of operation, algorithms in the abstract, or processes underlying the software.³

The distinction between the creative process and ownership of the finished product is particularly significant in the context of AI systems because much of the commercial value in AI often lies not merely in the written code itself, but in the model architecture, training methodology, optimisation processes, and datasets used to improve system performance. Consequently, ownership of an AI system cannot automatically be assumed simply because a business paid for or deployed the technology.

Ownership rights depend substantially on the manner in which the AI system was developed. Where an AI system is created by employees during the course of employment, ownership will, subject to the terms of the employment contract and any express intellectual property provisions, generally vest in the employee.⁴

Conversely, where development is outsourced to software developers, consultants, or technology vendors, ownership will primarily be governed by contractual terms. In the absence of clear intellectual property assignment clauses, the developer may retain proprietary rights in the AI system even where the commissioning business funded the development process.

Machine learning systems introduce further complexity. Unlike conventional software, machine learning models continuously evolve through training, retraining, and deployment on datasets. Nigerian law does not yet expressly address ownership of trained models, machine-generated improvements, or autonomously generated outputs. As a result, businesses must rely heavily on carefully drafted agreements to allocate ownership rights relating to pre-trained models, fine-tuned systems, derivative works, improvements, and downstream outputs, particularly where multiple parties contribute datasets, technical expertise, infrastructure, or cloud computing resources.

The Multi-Layered Nature of AI Ownership

Understanding ownership rights in AI systems requires recognising that ownership may exist across multiple layers of the technology, with different stakeholders owning distinct components.

A critical distinction must be drawn between the ownership of an AI system and the ownership of content generated through the use of that system. The former concerns the intellectual property rights subsisting in the software, models, datasets, and infrastructure that constitute the AI technology itself. The latter concerns the authorship and ownership of output generated through human interaction with the system. While these issues are often conflated, they raise fundamentally different legal questions and engage distinct areas of intellectual property law.

An AI system does not emerge in isolation; rather, it is the product of the efforts of engineers, researchers, data scientists, and software developers. The first layer of ownership therefore relates to the AI system itself, including the source code, model architecture, training systems, reinforcement learning processes, interfaces, datasets, and optimisation systems. It is these human-created elements that attract protection under intellectual property law.

Accordingly, ownership of an AI system is tied to several forms of intellectual property protection. For example, at this layer of ownership sits OpenAI, the developer and owner of ChatGPT. OpenAI’s proprietary rights arise from several distinct forms of intellectual property protection that are attached to different components of the system. The written software code, documentation, interface elements, training pipelines, and certain database structures are protected by copyright.

In addition to copyright protection, OpenAI benefits from trade secret protection in relation to proprietary components such as model weights, training methodologies, optimisation techniques, proprietary datasets, reinforcement learning processes, and safety mechanisms. Indeed, many AI companies rely more heavily on confidentiality obligations and trade secret protection than on copyright alone. There is also the trademark dimension of intellectual property protection. Names such as “ChatGPT” and “Gemini”, together with their associated logos, branding, and goodwill, are protected under trademark law.

Ownership and control are not determined solely by intellectual property law. Contractual arrangements also play a significant role in allocating rights, responsibilities, and commercial interests among the parties involved. Through licensing agreements and terms of use, AI developers are able to define how their systems may be accessed, deployed, modified, and commercialised. What follows is the interaction between users and AI systems. As AI tools become increasingly integrated into creative and commercial activities, attention inevitably shifts from ownership of the system itself to ownership of the outputs generated through its use

Ownership of AI-Generated Outputs

One of the most significant contemporary legal issues arising from AI adoption is the growing prevalence of AI-generated outputs. Particularly contentious is the question of ownership of content created using AI-driven systems.

Traditional copyright principles are premised on human authorship. Consequently, where content is autonomously generated by AI systems with minimal human involvement, questions arise as to whether copyright protection exists at all and, if so, who may claim ownership. Conversely, where a human user exercises sufficient creative direction, editorial judgment, or control over the AI system, such user may be able to assert authorship rights over the resulting output.

Across a range of creative industries, including publishing, film, advertising, music, and digital media, disputes concerning the ownership of AI-generated content are becoming increasingly commonplace. The emergence of AI-generated books, scripts, artworks, and musical compositions has reignited longstanding debates about originality, creativity, and authorship.

Traditional copyright principles generally contemplate human creators rather than machines. As a result, many legal scholars and policymakers continue to debate whether AI-generated works should attract copyright protection and, if so, should be recognised as the author. An increasingly accepted view is that where meaningful human input, oversight, and creative control exist, the human contributor may be entitled to claim copyright in the resulting work.

The growing concern regarding authorship and authenticity has also extended to the music industry. As AI-generated music becomes more prevalent, industry stakeholders continue to explore mechanisms for distinguishing human-created works from AI-generated content. Such developments reflect the increasing importance of transparency and attribution in the evolving digital creative economy.⁵

The Role of Data Protection in AI Development

Data protection law has a direct and material impact on AI systems that rely on personal data. Under the Nigeria Data Protection Act (NDPA) 2023, personal data used to train or operate AI systems must be processed lawfully, fairly, and transparently. Organisations must identify a lawful basis for processing, comply with purpose limitation, and implement appropriate technical and organisational measures to safeguard data.

AI systems that involve profiling or automated decision-making capable of producing legal or similarly significant effects on individuals attract heightened scrutiny. Organisations are required to ensure transparency around how such systems operate, provide meaningful information about the logic involved, and implement safeguards to protect data subjects’ rights. While the Act does not determine ownership of the components of AI systems, it constrains how data-driven AI systems may be designed, trained, and deployed, particularly in sensitive sectors such as finance, healthcare, and employment.

Regulatory oversight in this area rests primarily with the Nigeria Data Protection Commission (NDPC), whose approach increasingly emphasises accountability, risk assessment, and demonstrable compliance.

Liability and Accountability for AI Outputs

Beyond ownership concerns, liability remains one of the most critical legal issues surrounding the deployment of AI in Nigeria. Nigerian law does not recognise AI systems as legal persons capable of owning property, assuming legal obligations, or bearing liability independently. Accordingly, responsibility for AI-related harm ultimately rests with the natural or legal persons involved in the design, deployment, operation, or use of the system.

Where an AI system causes financial loss, discriminatory outcomes, reputational harm, or personal injury, liability will typically be assessed under existing legal principles relating to negligence, contract, consumer protection, and statutory obligations. Businesses deploying AI systems may therefore be exposed to liability where they fail to exercise reasonable care in the design, testing, monitoring, or oversight of such systems.

Liability allocation within commercial relationships will largely depend on contractual arrangements between end-users, developers, vendors, service providers, and deploying organisations. Properly drafted contracts should clearly address warranties, indemnities, limitations of liability, regulatory compliance obligations, data usage rights, and responsibility for system failures or unlawful outputs.

In the absence of clear contractual provisions allocating risks and liability, businesses and individuals will be vulnerable to the significant adverse consequences resulting from substantial exposure to AI-generated integration which they have limited technical control over.

The Nigeria Data Protection Act further reinforces accountability obligations by saddling data controllers with primary responsibility of ensuring the lawful processing of personal data, even where automated systems or third-party processors are involved.⁶ Consequently, where AI systems generate unlawful, biased, or discriminatory outcomes involving personal data, regulators are likely to hold the deploying organisation accountable rather than the technology itself.

Consumer protection risks are likely to arise where AI systems are deployed in products or services offered to the public. Misleading recommendations, inaccurate outputs, discriminatory profiling, or unsafe automated decisions may expose businesses to regulatory sanctions, reputational damage, and civil liability claims.

Nigeria’s Emerging AI Governance Landscape

At the policy level, Nigeria’s AI governance landscape is still in its formative phase, a phase where potential impact outweighs actual outcomes. The National Information Technology Development Agency (NITDA) continues to issue notes and provide guidance on digital innovation, emerging technologies, and responsible technology use, while the Nigeria Data Protection Commission (NDPC) provides regulatory oversight where AI systems intersect with personal data processing. Together, these institutions signal a growing three-pronged focus on transparency, accountability, and risk management in AI deployment.

While Nigeria has not yet enacted a dedicated AI law, global developments and regional policy trends suggest that more formal regulation is likely. Businesses operating in Nigeria should therefore anticipate increasing scrutiny of AI systems, particularly those deployed at scale or for high-risk use cases.

Key Considerations for Businesses

Businesses deploying AI systems in Nigeria should recognise that ownership and accountability are rarely straightforward. Ownership rights in AI systems, datasets, trained models, and generated outputs should be expressly addressed in employment agreements, consultancy arrangements, licensing agreements, SaaS contracts, and technology development documentation.

Businesses should also appreciate that copyright protection extends primarily to source code and not to the underlying ideas, logic, methodologies, or machine learning concepts underpinning the AI system. Consequently, organisations should complement copyright protections with robust trade secret protections, confidentiality obligations, access controls, and data governance measures.

As Nigeria’s technology ecosystem continues to evolve, especially against the backdrop of AI, organisations that proactively align their intellectual property strategy, contractual arrangements, data protection compliance obligations, and risk management frameworks will be better positioned to navigate the complex legal issues associated with AI ownership, deployment, and accountability.

Footnotes

1. Section 108, Copyright Act, 2022

2. Section 2(2) Copyright Act, 2022

3. Section 3, Copyright Act, 2022

4. Tonk A, Jain R and Sharma S (Ai at work: What employers must require, document, and enforce on confidentiality, IP and responsible AI use | International Bar Association, 21 April 2026) <https://www.ibanet.org/AI-at-work-what-employers-must-require> accessed 14 May 2026

5. https://www.cbsnews.com/news/spotify-verified-badge-ai-music/

6. Section 24(3) NDPA 2023

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.