NETHERLANDS Law and Practice Contributed by: Roderik Vrolijk, Rogier Raas, Ingrid Viertelhauzen and Maarten Weekenborg, Stibbe
MiFID II (as implemented in the Wft) and are directly subject to its algorithmic trading requirements. Fund managers operating under the AIFMD or the UCITS Directive fall outside the scope of MiFID II when performing fund management activities. Where an AIFM is authorised to provide additional MiFID ser - vices (such as individual portfolio management), those specific activities are subject to MiFID II. In practice, if a fund’s trades are transmitted to a broker for execu - tion, the algorithmic trading obligations attach to the executing broker rather than the fund manager. The business models differ in that dealers trade for their own profit from spreads and latency, while fund managers invest pooled third-party capital and are subject to AIFMD governance, delegation and inves - tor protection rules. Both are subject to (distinct) licensing requirements, to the MAR and DORA’s ICT resilience requirements. 7.4 Regulation of Programmers and Programming Programmers developing trading algorithms are not individually licensed or regulated. Regulatory account - ability rests entirely with the investment firm deploy - ing the algorithm. An investment firm must govern the full algorithm life cycle, ensure staff have adequate regulatory knowledge and maintain responsibility for third-party code. Engaging external developers may trigger a regulatory outsourcing framework, requiring due diligence, contractual safeguards and ongoing oversight. The Dutch remuneration rules, including the variable remuneration cap, apply to all individuals working under the responsibility of a Dutch-licensed financial enterprise.
fraud detection, and most expect usage to increase significantly. Although underwriting processes are not directly pre - scribed by regulation, they are substantially shaped by it. AI applications used for risk assessment and premium setting in life and health insurance are clas - sified as high-risk under the EU AI Act. This requires compliance with standards relating to risk manage - ment, data quality, human oversight, robustness and transparency. Supervisory objectives and standards apply regardless of the technology used, including AI. 8.2 Treatment of Different Types of Insurance Under the EU AI Act, AI used for setting premiums and assessing risk in life and health insurance is catego - rised as high risk. This triggers requirements relating to the following: • risk management; • data quality; • technical documentation; Specifically in the insurance sector, AI-based premium personalisation could undermine the principle of soli - darity, potentially rendering certain groups uninsur - able. The AFM and DNB have signalled continued supervi - sory focus on the use of AI in insurance, with particu - lar attention to model risk management, data quality, transparency of decision logic and fair treatment of policyholders. Insurers are expected to map their AI applications and ensure that automated processes remain consistent with the duty-of-care obligations under the Wft. • human oversight; • robustness; and • transparency.
8. Insurtech 8.1 Underwriting Processes
9. Regtech 9.1 Regulation of Regtech Providers
Insurtech participants are increasingly using AI in underwriting for risk assessment, setting premiums and detecting fraud. Insurers use AI for behavioural pricing, determining premiums based on customer behaviour, drawing on new internal and external data sources. Among Dutch insurers, AI is most common - ly used for chatbots, targeted online marketing and
Whether a regtech provider is subject to regulation depends on the nature of the services it provides. A regtech designation does not, of itself, trigger or exempt a firm from regulatory requirements.
577 CHAMBERS.COM
Powered by FlippingBook