BLOG
The Techno-Legal Framework to Govern Autonomous Entities
Nooriam is the governance layer between organisational autonomous system environments. It is techno-legal infrastructure that addresses the challenge of managing liability and attributing value from the operation of systems such as agentic AI.
AI Agents are now active participants in economic and social transactions, and their role is expanding at pace.[1] The protocols, such as MCP and A2A, governing agent identity and inter-agent communication are evolving alongside that expansion.
One of the open and recognised challenges with agents and autonomous systems, and not completely addressed by existing protocols, is their asymmetric ability to act autonomously and their inability to bear liability. This is a problem for all organisations seeking to deploy these systems outside of their organizational boundaries. It is a technical problem, but it has a significant legal component. Protocol-layer evolution does not produce, and cannot by itself generate, the techno-legal infrastructure needed to adequately and dynamically tether an AI Agent's technical operation to the legal person who bears responsibility for it.[2] Accordingly, the value derived from autonomous systems is discounted by the liability created by an unrestrained agent.
We call this the Legal Attribution Gap. It’s bimodal: it spans not only organisation risk in deploying these systems, and allocating value created by these systems as well. Nooriam has been architecting and building a solution to this problem for the last three years as part of its legally authenticated AI service. To iterate, without a techno-legal framework, the agentic systems being deployed at scale today operate in a legal and governance vacuum that compounds rather than manages organisational risk.
Legal systems in all jurisdictions account for both natural and non-natural people, such as corporations. The legal management of different parties’ behaviors is through incentives and disincentives levied upon the entity themselves, but the chain of responsibility is always tied to identifiable natural people. For example, a corporation itself may be punished with pecuniary penalties, ultimately the directors are responsible for the actions of the company and are the ones that will be held liable.
Under current legal systems and available technologies, AI Agents are unable to be enforced against in the same way as humans because AI Agents do not have bodies, status or feelings[3] that can be manipulated with sufficient punitive weight to reinforce lawful AI Agent performance.[4] This issue is partially solved if the legal systems decide that any deploying organization adopts liability on behalf of the agent or system.[5]
Nooriam directly addresses this. Nooriam involves a conjoined technical and legal technology stack designed for ready adoption with rigorous engineering (legal and technical) and mathematics. Nooriam’s cloud based infrastructure manifests in the middleware, the bridge between the cloud and the different applications, databases and operating systems of organisational environments.
Nooriam delivers the governance layer between organisations’ environments in a broad sense and cloud infrastructure. This means that Nooriam embeds legal authentication, verification and identity for organisations interacting with the surrounding world. Specifically, the platform provides legal infrastructure for AI and autonomous systems to safely interact between organisations. Join our commercial and institutional partners to contact us and find out more.
[1] International Data Corporation, 'Agentic AI to Dominate IT Budget Expansion Over Next Five Years, Exceeding 26% of Worldwide IT Spending, and $1.3 Trillion in 2029, According to IDC' (Media Release, prUS53765225, 26 August 2025) <https://my.idc.com/getdoc.jsp?containerId=prUS53765225 (https://my.idc.com/getdoc.jsp?containerId=prUS53765225)>.
[2] Zhu et. al. “Designing Meaningful Human Oversight in AI”, 24.08.2025.
[3] Although we are aware from informal discussions that data scientists are studying the biological structures in the human brain responsible for ethics, morality and shame to ascertain whether and if so how it might be possible to replicate those structures digitally in AI models.
[4] “Judicial training: a key to successful transformation of the judiciary”, Cape Town, South Africa, 2019; Published in (2021) 33(5) JOB.
[5] Qoine Pte Ltd v B2C2 Ltd [2020] SGCA(I) 02. The Qoine case is considered the leading common law authority on algorithmic trading liability. Two trading firms had each deployed autonomous algorithms to execute cryptocurrency trades. One algorithm submitted orders at approximately 250 times the prevailing market rate; the other accepted them automatically. When the error was discovered, Qoine sought to reverse the trades on the grounds of unilateral mistake. The Singapore Court of Appeal, by majority, rejected that argument. The majority held that the subjective state of mind relevant to a mistake in equity must be that of a natural or legal person, not of an algorithm, and attributed the relevant state of mind to the human operators who had programmed and deployed the accepting algorithm.