Section 196 ECCTA and the Case for Algorithmic Senior Management

Abstract

This paper critiques the 'individual' requirement in the UK's reform of corporate criminal liability. While the Economic Crime and Corporate Transparency Act 2023 (ECCTA) successfully dismantled the 'directing mind' doctrine, it replaced it with a human-centric 'Senior Manager' test that fails to account for the algorithmic delegation of management. By analysing the 2026 enforcement landscape and the proposed expansion of these rules via the Crime and Policing Bill 2025, this article argues for a functional re-interpretation of s.196. It proposes a 'Constructive Attribution' model where autonomous systems performing significant managerial roles are treated as de facto senior managers, thereby closing the 'responsibility gap' created by the black-box nature of modern corporate governance.

I. Introduction

For over a century, the English judiciary grappled with the "metaphysical problem" of corporate personhood. The traditional 'directing mind and will' doctrine, established in Lennard's Carrying Co Ltd v Asiatic Petroleum Co Ltd (1915), sought to solve this by identifying a human "ego" at the apex of the corporate hierarchy. However, as the Law Commission has correctly identified, this formalistic approach became a "shield for large, decentralized organizations." The enactment of the ECCTA 2023 introduced Section 196, replacing the restrictive 'directing mind' test with a functional 'Senior Manager' test. Furthermore, the Crime and Policing Bill 2025 seeks to universalize this attribution model across the entire criminal code, moving beyond economic crime to encompass offences ranging from corporate manslaughter to environmental breaches. Yet, this legislative "modernization" suffers from a profound biological myopia. While Section 196 successfully pivoted from hierarchy to function, it remains tethered to the "individual"—a natural person—as the sole vessel of corporate intent. In the contemporary operational reality of 2026, this human-centricity creates a systemic "Responsibility Gap". As global firms increasingly delegate "significant roles" to autonomous algorithmic systems, the "manager" is no longer a person, but a process. This article argues that the current statutory framework is ill-equipped to handle the rise of Algorithmic Senior Management.

II. The Responsibility Gap: Algorithmic Opacity and the Failure of Attribution

The fundamental challenge in 2026 is that the Serious Fraud Office's (SFO) Guidance on Evaluating a Corporate Compliance Programme (2025) now mandates that firms demonstrate "dynamic, data-led" oversight. Yet, the very tools firms utilize to satisfy this requirement—AI-driven AML monitors and automated trading systems—frequently operate as "black boxes" that frustrate the Section 196 ECCTA attribution test.

A. The "Individual" Constraint in a Post-Human Management Tier

Section 196(4) of the ECCTA specifically defines a senior manager as an "individual" (a natural person). However, as the UK Jurisdiction Taskforce recently affirmed, English law remains steadfast that AI lacks legal personality and cannot be held responsible for criminal harm. Consequently, when a machine-learning AML filter autonomously "decides" to deprioritize suspicious activity alerts to maintain transaction liquidity, the SFO faces an evidentiary wall. If no single "individual" in the senior management tier possessed the requisite mens rea—because the decision was an emergent property of the software's neural weights rather than a human directive—the corporation may remain technically innocent of the substantive offence. This creates the systemic "Responsibility Gap", where criminal harm occurs but the causal link to a human manager is too attenuated to satisfy the high threshold of criminal attribution.

B. The Architect's Liability: Significant Role or Significant Software?

A novel legal battleground concerns the Chief Technology Officer (CTO). Under the functional test, these individuals arguably play a "significant role" in managing a "substantial part" of the firm's activities. However, attributing a crime to the organization via the CTO requires proving the CTO themselves committed the offence. In cases of "algorithmic drift," where the system behaves in ways the architect did not foresee, the CTO lacks the specific intent required for offences like fraud. Without a "guilty individual," the s.196 mechanism collapses. As a result, the SFO is increasingly forced to rely on the Failure to Prevent Fraud (s.199) backstop—a strict liability "omission" offence—rather than prosecuting the substantive crime.

 

C. Emergent Intent: Towards Constructive Attribution

To resolve this, this article proposes a shift from Natural Intent to Constructive Intent. Just as the "identification doctrine" once looked for a "directing mind," the 2026 landscape demands that the law identifies a "Directing Algorithm." If a firm grants an autonomous system the authority to manage a "substantial part" of its business, it has effectively delegated its "senior management" function to that code. Under this theory, the "intent" of the algorithm should be legally constructed as the intent of the organization, bypassing the biological requirement currently tethering s.196 to natural persons.

 

III. The Reform Proposal: Toward a "Technologically Neutral" Attribution Model

As the Crime and Policing Bill 2025 progresses through Parliament, it seeks to universalize the "Senior Manager" test for all criminal offences. However, by failing to modernize the definition of "individual," the Bill risks codifying a massive loophole for automated corporations. To ensure the rule of law remains effective, this article proposes a targeted amendment to Section 196 of the ECCTA.

Redefining the "Senior Manager": The current definition in Section 196(4) relies on biological status. I propose a Functional Equivalency Clause. The statutory definition should be amended to include not just natural persons, but any "Autonomous Algorithmic Agent" that performs a significant role in decision-making.

Proposed Statutory Language: Section 196(4A): For the purposes of this section, a 'senior manager' shall include an autonomous algorithmic agent where such agent performs a function which, if performed by a natural person, would satisfy the criteria in subsection (4)(a) or (b).

The "Attribution of Output" Theory: Critics may argue that an algorithm cannot possess mens rea. However, this proposal adopts a Constructive Intent model. If a corporation grants "significant" operational authority to an AI—such as a system that autonomously executes trades—the intent of the system is found in its objective operational logic. The law should treat the algorithm's output as the corporation's intent if the harm was a foreseeable consequence of the algorithm's optimization parameters.

The "Designated Architect" Safeguard: To avoid arbitrary liability, the law should establish a rebuttable presumption that the Senior Manager (Human) who signed off on the deployment of the AI is the "nexus" for attribution, regardless of whether they understood the specific criminal outcome.

IV. Conclusion

The shift from the 'directing mind' to the 'senior manager' was a necessary evolution for the 20th-century firm. However, as we navigate the complexities of 2026, the law must recognize that the "mind" of the corporation is increasingly digital. By adopting a technologically neutral definition of management, the UK can lead the global effort to ensure that corporate criminal liability is not merely a "biological" concept, but a functional reality that holds even the most automated entities to the standards of the rule of law.