What does FSRA Digital Investment Management guidance say about AI in banking?

Published Jul 15, 2019 · Last reviewed Nov 19, 2026

The FSRA's Supplementary Guidance on Digital Investment Management, first issued on 15 July 2019 for robo-advisers in Abu Dhabi Global Market, sets out how the FSRA expects firms to govern the algorithms behind automated investment services. The current version (VER02.061125) defines algorithm-based tools to include 'modern developments in artificial intelligence (AI) such as generative AI tools, large language models, machine learning' and names risks such as model poisoning, model drift and adversarial manipulation. It expects board and senior management oversight of the algorithm's design, performance, deployment and security, outcomes that are explainable, traceable and repeatable, regular testing, the ability to stop or change the algorithm, and ongoing monitoring for inaccuracy and bias.

OFFICIAL TEXT: assets.adgm.com ↗ · FINAL · UAE (CBUAE / DFSA / FSRA)

Checked 2026-11-19 · UAE (CBUAE / DFSA / FSRA), assets.adgm.com
DocumentFSRA Digital Investment Management guidance — Supplementary Guidance: Digital Investment Management (FSRA, Abu Dhabi Global Market)
Issued byUnited Arab Emirates financial regulators: the Central Bank of the UAE (CBUAE), the Capital Market Authority (formerly the Securities and Commodities Authority), the Dubai Financial Services Authority (DFSA) of the Dubai International Financial Centre and the Financial Services Regulatory Authority (FSRA) of Abu Dhabi Global Market, with the DIFC Authority and its Commissioner of Data Protection for Data Protection Regulation 10
TypeGuidance
StatusFinal
PublishedJul 15, 2019
Applies toApplicants for and holders of an FSRA Financial Services Permission in Abu Dhabi Global Market that provide digital investment management (robo-advice), fully digital or hybrid, including advising, managing assets and arranging deals. Guidance issued under section 15(2) of the Financial Services and Markets Regulations 2015: the FSRA is not bound by it, but it explains how existing authorisation rules apply
Official sourceassets.adgm.com ↗
Use casesTrading & capital markets · Model risk management · Generative & agentic AI · AI governance (general) · Customer-facing chatbots

What are the key points of FSRA Digital Investment Management guidance?

  • Scope: investment management services using algorithm-based tools with limited or optional human interaction, fully digital or hybrid; firms that only supply the technology are not Digital Investment Managers and need no permission.
  • AI: a footnote extends algorithm-based tools beyond deterministic or rule-based logic to AI, including generative AI tools, large language models and machine learning.
  • Risks named (4.2): model poisoning, model drift and adversarial manipulation; the FSRA also points firms to its IT Risk Management Guidance and to the 2021 Guidelines for Financial Institutions adopting Enabling Technologies.
  • Algorithm governance (4.6): internal governance giving the Board and Senior Management 'robust oversight and control over the design, performance, deployment and security of the algorithm', with clearly defined roles for everyone who oversees it.
  • Development and testing (4.7(a) and (b)): qualified staff, documentation of the decision tree or logic so that outcomes are 'explainable, traceable and repeatable', attention to behavioural biases in client questionnaires, regular testing, and a possible third-party audit for complex models.
  • Maintenance (4.7(c)): protect the integrity of the source code, keep the ability to stop or modify the model, detect and report programming errors and unexpected outcomes, and keep contingency plans for model failure or outage.
  • Monitoring (4.7(d)): ongoing monitoring and reviews to manage 'the risks of inaccuracy, bias or exception', with periodic Board and Senior Management review of the governance structure.
  • Disclosure: clients are told how the algorithm is relied on, its key assumptions and limitations, the degree of human involvement, and when the algorithm may fail or be halted or materially adjusted.

What did FSRA Digital Investment Management guidance change for banks?

The FSRA introduced the framework on 15 July 2019 with requirements for human oversight of the algorithm model, protection against behavioural biases, safeguards for the model's integrity and explainable, traceable and repeatable outcomes. The current version, labelled VER02.061125, carries those algorithm-governance expectations over to AI, including generative AI and large language models, and names AI-specific risks such as model poisoning and model drift.

Does the ADGM FSRA have AI rules for robo-advisers?

It has guidance. The Supplementary Guidance on Digital Investment Management explains how the FSRA's existing rules apply to firms that provide investment management through algorithm-based tools, which its current version defines to include AI such as generative AI tools, large language models and machine learning. The FSRA states it is not bound by the guidance and may impose other conditions for additional risks.

What does the FSRA expect from firms whose algorithms use AI?

Board and Senior Management oversight of the algorithm's design, performance, deployment and security, qualified staff, documentation so outcomes are explainable, traceable and repeatable, regular testing and possibly a third-party audit for complex models, safeguards for the source code, the ability to stop or modify the model, and ongoing monitoring of inaccuracy, bias or exception.

DateDocumentStatus
Nov 2, 2026DFSA AI Survey 2026 — New DFSA AI Survey 2026 findings: AI adoption among DIFC financial services firms has more than doubled since 2024Final
Oct 30, 2026DFSA frontier and agentic AI cyber paper — Frontier and agentic AI: cyber resilience in the financial sector (DFSA insights paper with C:>DIR)Final
Jun 4, 2026DFSA Dear SEO letter on AI risk management — DFSA regulatory expectations on artificial intelligence risk management in the DIFC (letter to Senior Executive Officers of Authorised Firms)Final
Feb 11, 2026CBUAE AI and Machine Learning Guidance Note — Guidance Note on Consumer Protection and the Responsible Adoption and Use of Artificial Intelligence and Machine Learning by Licensed Financial Institutions in the U.A.E.Final
Nov 12, 2025DFSA Artificial Intelligence Survey 2025 — DFSA Artificial Intelligence Survey 2025: An overview of regulatory insights and trends in financial services in the DIFCFinal
Sep 1, 2023DIFC Data Protection Regulation 10 — DIFC Data Protection Regulations, Regulation 10: Personal Data Processed Through Autonomous and Semi-Autonomous SystemsIn force

Follow every document these regulators publish

when one of these regulators moves, the next morning's brief says so · six sourced stories · 7 am PT · free

plus every regulation, bank and agent page update, the morning after · leave any morning