Book a demo
Topics · AI-DLC

Banking AI-DLC DORA | Moring

For banking compliance leaders overseeing AI-DLC integrations, digital operational resilience requires aligning code generation and automated systems with DORA frameworks. Institutions must establish rigorous change management, risk governance, and third-party oversight to ensure that AI deployments in financial technology environments remain fully secure, auditable, and resilient.

On this page
  1. Short answer
  2. How it works
  3. FAQs
  4. Related reading
  5. Sources

Short answer

For banking compliance leaders overseeing AI-DLC integrations, digital operational resilience requires aligning code generation and automated systems with DORA frameworks. Institutions must establish rigorous change management, risk governance, and third-party oversight to ensure that AI deployments in financial technology environments remain fully secure, auditable, and resilient.

How it works

Integrating artificial intelligence into regulated banking engineering organizations introduces complex operational hurdles under the Digital Operational Resilience Act, which has applied since January 2025. Where AI-generated code results in a change to an information and communication technology system, that change falls within the institution's existing DORA change-management framework, encompassing risk assessment, testing, approval, and controlled deployment. DORA places ultimate responsibility for ICT risk squarely with the management body, while requiring risk-based change management and governance processes throughout the entire ICT environment.

Controlling AI-DLC pipelines requires strict operational parameters. Traditional ICT risk management assumes deterministic systems that behave exactly as coded, but AI and automated agents introduce probabilistic behavior where identical inputs may produce different outputs. These systems learn, adapt, and chain decisions together without human intervention, expanding the enterprise asset inventory. Every AI agent used in banking operations is an ICT asset that requires documented purposes, data access parameters, decision authorities, model versions, and dependency mapping.

Regulatory expectations and industry guidelines emphasize that financial institutions must ensure AI-based systems are consistently governed, secured, and monitored within established DORA-compliance frameworks. According to regulatory guidance, institutions must adopt a management-approved AI strategy, define clear responsibilities, build internal AI competencies, and ensure interdisciplinary collaboration, particularly where AI supports critical or important functions. Furthermore, firms must apply robust development, change management, and documentation standards to in-house AI developments, paying strict attention to open-source components and AI-assisted code generation that might introduce hidden dependencies or vulnerabilities.

A robust practitioner checklist for banking technology leaders implementing AI-DLC under DORA must include: establishing comprehensive asset inventories for every AI model and agent; extending DORA-compliant ICT risk management frameworks to cover identification, protection, detection, response, and recovery; incorporating AI agent attack surfaces into regular resilience testing and threat-led penetration testing; enforcing contractual and oversight controls for all third-party AI and machine learning vendors; and ensuring that management-level governance actively oversees model risk alongside operational resilience.

Isolated code generation in banking software development frequently fails because models lack necessary organizational context. Without guidance, an AI model might produce code that uses a library conflicting with approved dependencies, implements authentication patterns violating secure coding standards, creates data structures prohibited by data classification policies, or duplicates existing enterprise services. These failures underline the necessity of integrating architecture decision records, API specifications, approved dependency manifests, secure coding guidelines, and data classification policies directly into the AI-DLC workflow.

To begin addressing DORA requirements within the AI development lifecycle, banking technology teams should first map all existing AI-assisted coding tools and autonomous agents to the enterprise ICT asset inventory. Leaders must then establish cross-functional collaboration between engineering, compliance, and risk management units to align AI strategy with technology roadmaps, ensuring that management bodies retain ultimate accountability for AI-related ICT risk oversight while deploying proportionate controls across the entire AI lifecycle.

Frequently asked questions

How does DORA impact AI-assisted code generation within banking engineering?

Under DORA requirements, any AI-generated code modification to a banking ICT system falls directly under established change-management frameworks. This mandates rigorous risk assessment, thorough testing, formal approvals, and controlled deployment procedures, ensuring that banking technology teams maintain strict oversight over automated software development workflows.

What specific ICT risks do autonomous AI agents introduce to banking operations?

Autonomous AI agents introduce probabilistic behavior where identical inputs can yield varying outputs, challenging deterministic banking ICT risk models. Because these agents learn, adapt, and chain decisions independently, banking institutions must expand their asset inventories and maintain comprehensive documentation regarding agent data access, decision authority, and model dependencies.

What are the core governance requirements for banking institutions deploying artificial intelligence?

Banking institutions must adopt a management-approved AI strategy that aligns with technology roadmaps for ICT resources and capacity. Within the AI-DLC, leadership must establish internal control frameworks, build interdisciplinary competencies, and ensure that management bodies retain ultimate accountability for AI-related ICT risk oversight and resilience testing.

Why do isolated AI code generation tools fail in enterprise banking environments?

Isolated code generation tools lack organizational context, often producing banking application code that violates secure coding standards, uses unapproved dependency libraries, or breaches internal data classification policies. Providing proper organizational context through architecture records and approved manifests is essential for banking compliance.

Sources

  1. - federalreserve.gov
  2. - federalreserve.gov
  3. - ithandbook.ffiec.gov
  4. - artificialintelligenceact.eu
  5. - eur-lex.europa.eu