Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM65887
Volume:
10 |
7 |
Jul 2026
Pages:
1-4
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.
Management
Swami Ramanand Teerth Marathwada University, Nanded
Management
Peoples College, Nanded
As the financial sector increasingly adopts artificial intelligence (AI), the development of governance frameworks has been lagging behind, leaving a very wide gap between the deployment of AI and the implication of its risks. The research is based on the 2025 EMEA Model Risk Management Survey. It covered survey of 87 banks and 49 insurance companies across Europe, the Middle East and South Africa in order to explore patterns of AI adoption, implementation challenges, and governance issues in financial institutions. Transparency and explainability continue to be top cited challenges, along with skills gaps and regulatory uncertainty. To fill these gaps, this paper suggests the framework of System-Level Intelligence Assurance, which is based on four pillars of architecture, shared memory, continuous validation loop, confidence governance, and auditable evidence chain. It is an extension to the concept of Model Risk Management that introduces agentic AI systems and replaces the old governance model of paperwork with one that is ongoing and built into the architecture. The results provide valuable insights for financial institutions, regulators, and risk practitioners aiming to safely scale AI while ensuring strong oversight and accountability.
KEYWORDS
The 2025 EMEA MRM Survey reveals a financial services industry is rapidly embracing AI. But the industry is struggling to govern it effectively. The adoption rates have surge with 67% of banks and 57% of insurers now using AI/ML models, however a significant governance gap persists. Only 29% of banks and 15% of insurers consider their MRM frameworks adequate for AI oversight. The main hurdles include transparency challenges, skills shortages, and regulatory uncertainty. To eliminate this gap, a proposal of System-Level Intelligence Assurance—a paradigm shift from model-level to system-level governance is been suggested. This is occupied in four pillars: § Shared Memory – Instead of fragmenting in centralized structures, it allows intelligence to build up across agents. § Continuous Validation Loops – substitutes runtime oversight for recurring reviews. § Confidence Governance – Ensures trust is earned through displayed behavior, not assumptions. § Auditable Evidence Chains – Makes every decision verifiable, not just fluent. Banks, insurers, and regulators are guided to not use governance challenges as a reason to slow the adoption of AI. Instead, they should treat the challenges as a reason to upgrade in the way they govern AI. System-level assurance builds trust directly into the technology itself which will make it continuous and measurable.
1. Fritz-Morgenthal S, Hein B and Papen Brock J (2022) Financial Risk Management and Explainable, Trustworthy, Responsible AI. Front. Artif. Intell. 5:779799. Doi: 10.3389/frai.2022.779799 2. Deloitte. (2025). EMEA Model Risk Management Survey 2025 (Banks and Insurers report). 3. Kurshan, E., Balch, T., & Byrd, D. (2025). The Agentic Regulator: Risks for AI in Finance and a Proposed Agent-based Framework for Governance. arXiv. 4. Deloitte India. (2026). Building Trust in Agentic AI: TRACE Framework for Policy-Driven Multi-Agent System Design. Zenodo. 5. Djeffal, C., Siewert, M. B., & Wurster, S. (2022). Role of the state and responsibility in governing artificial intelligence: a comparative analysis of AI strategies. Journal of European Public Policy, 29(11), 1799–1821. https://doi.org/10.1080/13501763.2022.2094987
Publication Date
7/7/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
1
Total views since publication
PDF Downloads
9
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...