Ai Governance In European Investment Management

With the increasing adoption of artificial intelligence by investment managers in Europe, some questions arise that go beyond predictive performance. Investment firms will need to consider not only whether an AI system is caught by the scope of the EU AI Act, but also how its use interacts with existing governance, conduct, algorithmic trading and operational resilience requirements. The paper discusses regulatory constraints that apply to AI systems used for investment research, portfolio construction, risk management and algorithmic trading. The study is founded on a systematic documentary analysis of the EU AI Act, MiFID II, RTS 6, DORA and recent ESMA publications in order to distinguish between formal legal obligations and more general supervisory expectations and model governance. This demonstrates that portfolio selection models are not themselves high-risk systems under the AI Act and that explainability is not a uniform legal requirement for all investment models. But transparency and auditability still matter for internal validation, human oversight, model-risk management and supervisory accountability. 'This paper introduces a proportionate governance framework with three dimensions: the role of the AI system in the investment process, the degree of autonomy of the AI system and the potential impact on clients, firms and market integrity'. The framework connects these dimensions to differentiated requirements for documentation, validation, explainability, monitoring, human intervention, and third party oversight. The analysis will provide an operational framework for European asset managers to balance technological innovation with accountability and regulatory compliance.