AI & Funds #6 –
Conflicts of Interest

The purpose of disclosing conflicts of interest in the investment services industry is to ensure maximum transparency for the investor.

A common conflict of interest is the situation when the fund’s investment manager owns voting shares in the fund and appoints a director who is already heavily involved in the investment manager’s structure. Consequently, the appointed director might be conflicted if one is required to take a decision against the investment manager. To mitigate this conflict of interest, the fund would apply its existent policy on conflicts of interest which would stipulate that the conflict be disclosed during board meetings and the director refrains from voting in matters relating to the conflict of interest.

Such mitigation might seem common sense in practising good corporate governance. However, Directive 2014/91/EU of the European Parliament and of the Council of 23 July 2014 (“UCITS V”) amending Directive 2009/65/EC on the coordination of laws, regulations and administrative provisions relating to UCITS (“UCITS Directive”) which regulates UCITS funds and Directive 2011/61/EU of the European Parliament and of the Council of 8 June 2011 on Alternative Investment Fund Managers (“AIFMD”) which regulates Alternative Investment Funds (“AIFs”) and AIF Managers contain very clear rules on conflicts of interest which go beyond the disclosure which is expected from non-licensed entities.

Nevertheless, neither the UCITS Directive nor the AIFMD provide for rules to mitigate conflicts of interest which arise from systems of Artificial Intelligence (“AI”). The use of AI systems in UCITS and AIFs can give rise to conflicts of interest which can potentially breach the regulator’s conduct supervision guidelines if not properly disclosed.  A few examples are:

• During the selection of investment opportunities, it is possible that investment managers may utilise AI to identify and make investment decisions. Nonetheless, the algorithms implemented may prioritise investments that generate higher fees for the investment manager, rather than those that offer the best returns for the investors, leading to a conflict of interest.
• Data accuracy is essential for AI models to function efficiently. Service providers may use incomplete or inaccurate data, or utilise data that has not been properly anonymized, leading to potential biased or incorrect investment decisions, or unlawfully exposing the investor’s investment preferences.
• Investment managers may employ proprietary AI algorithms to make investment decisions, which may create a conflict of interest as the investment manager may have a financial interest in keeping the algorithm confidential, even if it is not in the best interests of the investors.
• AI may be used to manage risk in investment portfolios by investment managers, and the algorithms used may prioritise the reduction of risk for the investment manager rather than the investor. This may result in investments that are less risky, but offer lower returns, creating a conflict of interest.
• The lack of disclosure of the use of AI in investment decisions may create a conflict of interest, as investors may not understand how their money is being invested, making it difficult for them to determine if the investment manager is making decisions in their best interest in terms of the offering memorandum.

To understand how AI-related conflicts of interest might be tackled, one needs to look at the existing approaches by the EU regulator in non-AI-related matters. Taking Article 25 of the UCITS directive as an example, “A depositary shall not carry out activities with regard to the UCITS or the management company on behalf of the UCITS that may create conflicts of interest between the UCITS, the investors in the UCITS, the management company and itself, unless the depositary has functionally and hierarchically separated the performance of its depositary tasks from its other potentially conflicting tasks, and the potential conflicts of interest are properly identified, managed, monitored and disclosed to the investors of the UCITS.”

If one were to apply the UCITS’s approach of mitigating conflict of interest to the application of AI, the use of AI would not be prohibited as long as it is disclosed to the investors. Having said that, this would require a higher level of disclosure than that required by the ‘Amendments adopted by the European Parliament on 14 June 2023 on the proposal for a regulation of the European Parliament and of the Council on laying down harmonised rules on artificial intelligence (Artificial Intelligence Act) and amending certain Union legislative acts (COM(2021)0206 – C9-0146/2021 – 2021/0106(COD)’ (the “draft EU AI Act”)¹. The reason being that the draft EU AI Act contains transparency obligations which would only apply to high-risk AI systems and limited-risk AI system as explained in a previous insight briefing in this series.

Eventually, as a minimum, the disclosure of AI in EU asset management would likely require the standard of disclosure required for AIFs rather than UCITS as the former triggers less stringent reporting obligations. Whilst one does not envisage that a separate policy on conflicts of interest from AI would be required for each licence-holder, existing policies might need to be amended to reflect AI usage in the same way as these policies had to be amended in relation to sustainability in the first part of this decade.

In fact, the fifth preamble of Commission Delegated Regulation (EU) 2021/1255 of 21 April 2021 amending Delegated Regulation (EU) No 231/2013 as regards the sustainability risks and sustainability factors to be taken into account by Alternative Investment Fund Managers says that:

“To maintain a high standard of investor protection, AIFMs should, when identifying the types of conflicts of interest the existence of which may damage the interests of an AIF, include conflicts of interest that may arise as a result of the integration of sustainability risks in their processes, systems and internal controls. Those conflicts may include conflicts arising from remuneration or personal transactions of relevant staff, conflicts of interest that could give rise to greenwashing, mis-selling or misrepresentation of investment strategies and conflicts of interests between different AIFs managed by the same AIFM.”²

Subsequently, Article 1(4) of Commission Delegated Regulation (EU) 2021/1255 amended Article 30 of Commission Delegated Regulation (EU) No 231/2013 which now reads: ‘AIFMs shall ensure that when identifying the types of conflicts of interest…they shall include those types of conflicts of interest that may arise as a result of the integration of sustainability risks in their processes, systems and internal controls.’

The draft EU AI Act includes provisions for fines and other penalties for non-compliance with the AI legislation which are uniformly calculated across the internal market. Moreover, in June 2023, the EU Parliament had proposed that the draft EU AI Act should “establish an AI Office, a new EU body to support the harmonised application of the AI act, [to] provide guidance and coordinate joint cross-border investigations.”³ Such proposal was included in the draft EU AI Act’s new⁴ text which was adopted by the EU Parliament’s plenary vote on the 14^th of June 2023.

This is in contrast with the enforcement of UCITS and AIFs which primarily relies on administrative penalties by local financial regulators.  The divergent approach to enforcement further confirms the existing notion that the draft EU AI Act will be more akin to a lex generalis which can be further elaborated either through directives (such as the draft AI liability directive⁵) or through EU Commission delegated regulations and regulatory technical standards.