The governance and accountability gap in AI adoption

The governance and accountability gap in AI adoption

Arpinder Singh - Emerging Markets Leader, Forensic & Integrity Services, EY India
Arpinder Singh

Emerging Markets Leader, Forensic & Integrity Services, EY India

5 minute read 23 Jul 2026

AI adoption is outpacing governance, creating accountability and traceability gaps, making structured frameworks such as ISO 42001 critical.

In brief

  • AI use is rapidly expanding into business-critical decisions, but governance frameworks are not keeping pace, leading to gaps in oversight and accountability.
  • A lack of traceability and documentation makes it difficult to explain, audit or investigate AI-driven outcomes, especially in risk or fraud scenarios.
  • Frameworks such as ISO 42001 provide structured governance, enabling lifecycle oversight, risk management and defensible, transparent AI-driven decisions.

Artificial intelligence (AI) is no longer limited to experimentation or innovation labs. It is increasingly being incorporated into customer engagement, operational workflows, risk assessment and decision-making processes across organizations. As adoption accelerates, AI is moving from a supportive technology to a core business function that directly influences outcomes.

However, AI governance structures have not always evolved at the same pace, particularly with respect to governance frameworks and compliance requirements. While organizations continue to invest in AI capabilities, there is often less clarity on how these systems are governed once deployed under an AI Management System. In many cases, governance exists at a policy level but is not consistently placed into operational practice. This creates a gap between AI deployment and effective AI oversight.

This gap becomes more visible as AI systems begin influencing decisions with financial, operational and regulatory consequences, increasing the need for accountability and risk management. Questions around accountability often become more complex in such environments. It may not always be clear who is responsible for an AI-driven decision, how that decision was made, or what controls were in place to support its reliability. As a result, governance challenges are not limited to system design but extend to ongoing use and monitoring.

Ethical considerations remain an important underlying aspect of AI governance, particularly when it comes to human oversight in AI. These systems increasingly influence decisions that affect individuals, customers and financial outcomes. Beyond compliance, organizations are expected to demonstrate that AI-driven processes are fair, responsible and free from unintended harm. However, in practice, ethical objectives should be supported through structured governance and control mechanisms to become effectively operationalized. Without these mechanisms, ethical principles often remain conceptual rather than enforceable.

The issue becomes more critical when AI-driven decisions are challenged, produce unexpected outcomes or are associated with fraud or misconduct, requiring the implementation of strong AI crisis management and incident response practices. In such situations, organizations are expected to reconstruct the decision-making process and demonstrate that appropriate controls and monitoring were in place. However, this is not always straightforward. In many deployments, documentation is limited, and decision pathways are not designed to be easily reconstructed after the fact.

From a forensic perspective, this creates a practical challenge and highlights the need for forensic readiness. AI systems often rely on complex models, multiple data inputs and layered decision processes. Once deployed, it can be difficult to establish how a specific outcome was generated. In the absence of structured governance and AI audit trail capabilities, even routine AI-enabled decisions can become difficult to investigate retrospectively. This limits the ability to determine whether an issue arose from system design, data quality issues or deliberate manipulation. It also increases the time and complexity associated with regulatory or internal investigations.

EY Document Anomaly & Transaction Analytics - Document verification and fraud detection

EY Document Anomaly & Transaction Analytics offers multi-level checks and advanced statistical algorithms for accurate document verification.

Know more

EY Trusted Verification - Employee Background Check and Verification

Trusted Verification at EY supports organizations BGV process with employee background verification and screening services for informed hiring decisions.

Know more

This is where structured governance frameworks such as ISO 42001 become relevant in enabling AI lifecycle governance. ISO 42001 is an international standard that provides a structured approach to managing AI systems throughout their lifecycle. It is designed as an AI Management System (AIMS), similar in concept to other ISO management standards and focuses on establishing governance rather than defining technical AI design.
 

The framework is built around integrating AI governance into organizational processes. It addresses areas such as defining roles and responsibilities for AI systems, adopting structured risk management practices, maintaining documentation of AI lifecycle decisions and establishing mechanisms for monitoring AI performance over time through controls and monitoring. It also encourages organizations to address risks related to data quality, model behavior and human oversight in AI as part of a unified governance approach.
 

Importantly, ISO 42001 places emphasis on decision traceability throughout the AI lifecycle. This means organizations are encouraged to maintain sufficient records to understand how AI systems are developed, deployed and used in decision-making. In practice, this becomes essential when decisions need to be reviewed or explained, particularly in situations involving disputes, investigations or regulatory scrutiny.
 

While ISO 42001 does not eliminate the risks associated with AI, it provides a consistent governance framework for managing them and supporting compliance. As AI becomes embedded in higher-impact business decisions, such structures become increasingly important in enabling organizations to explain, review and defend outcomes when required.
 

The challenge is no longer simply about adopting AI but about governing it in a manner that can withstand regulatory scrutiny through responsible AI governance. As reliance on AI systems increases, the ability to explain and reconstruct AI-driven decisions will become central to organizational accountability and trust. Frameworks such as ISO 42001 represent an important step toward developing that capability.

Learn more about AI governance and accountability for the responsible adoption of AI

Summary

AI is increasingly embedded in core business functions and influences key decisions across organizations. However, governance and accountability frameworks have not evolved at the same pace, creating gaps in oversight, traceability and control. This makes it difficult to clearly assign responsibility, explain outcomes or investigate issues, particularly in cases involving risk, fraud or regulatory scrutiny. Ethical intent alone is insufficient without structured implementation. Frameworks such as ISO 42001 provide a lifecycle-based governance approach, emphasizing documentation, monitoring and risk management. As organizations become more reliant on AI, the ability to transparently explain and defend AI-driven decisions will be essential to maintaining trust and accountability.

How EY can help

Related articles

How AI governance supports confident and responsible growth

AI governance is now a boardroom priority. Find out why enterprises should embed accountability, oversight and trust in AI-driven decisions.

Arpinder Singh

Why document integrity checks fall short and how AI changes the equation

Learn how AI-powered document verification improves integrity checks, detects fraud at scale, and strengthens enterprise trust.

Anurag Kashyap

India Inc. is harnessing AI as a real-time fraud watchdog

AI can play an enabling role in real-time fraud detection for India’s financial ecosystem by identifying patterns and flagging anomalies.

Arpinder Singh

About this article

Arpinder Singh - Emerging Markets Leader, Forensic & Integrity Services, EY India
Arpinder Singh
Emerging Markets Leader, Forensic & Integrity Services, EY India
Leading forensic accountant and veteran expert witness. Advising global clients on compliance, anti-corruption and corporate governance. Disruptive thinker. Technology aficionado. Author.
Related topics
Forensics
AI
Technology
Digital
Risk

EY refers to the global organization, and may refer to one or more, of the member firms of Ernst & Young Global Limited, each of which is a separate legal entity. Ernst & Young Global Limited, a UK company limited by guarantee, does not provide services to clients.