Listen now or read the full briefing below
Overview
A regular briefing for the alternative asset management industry.
Self-evidently, artificial intelligence (AI) and machine learning (ML) offer new opportunities for alternative asset managers – with many firms already using them to help with a range of key tasks, including deal analytics, due diligence and portfolio company monitoring. Proprietary technologies sit alongside ready-made tools like Microsoft Co-pilot to increase efficiency and effectiveness. But deployment of AI also brings challenges – and therefore the attention of regulators. In the UK, the financial regulators have fired the starting gun, setting out a strategic approach to managing AI integration in financial services.
The UK's Financial Conduct Authority (FCA) issued an AI update in April which explores the evolution of AI in financial markets and sets out the FCA's approach. (Our detailed note on the FCA paper is available here.)
In essence, the FCA is not proposing AI-specific rules (at least, not for now), but emphasises that a range of existing FCA rules are all potentially engaged when a firm is using and/or procuring AI. There may not be AI-specific FCA regulation, but there is now FCA regulation of AI.
The FCA's report notes the rise of AI technologies, as increasingly sophisticated algorithmic models permeate all sectors of the economy (including law). It welcomes AI's ability to enhance decision-making processes and generate significant cost savings. The FCA suggests AI and ML could fundamentally reshape finance, turning data into a significant asset and driving a shift towards more personalised services.
However, the FCA recognises AI as a powerful double-edged sword. To counter risks, the FCA is explicit that firms need to maintain appropriate governance frameworks and controls. Risks, primarily relating to opaque decision-making processes and data bias, could undermine trust in AI systems and raise ethical questions. AI's heavy reliance on data also increases security hazards, with potential breaches carrying heavy consequences in financial markets.
Moreover, the FCA notes the need for human understanding and oversight of AI models, referring to the need for appropriate transparency. While such tools can outperform more traditional models, their inability to explain their decision-making process creates a trade-off between performance and explainability.
Many firms use third-party vendors for AI and machine learning applications. Although this may aid innovation, the FCA emphasises the need for firms to possess sufficient expertise to ensure comprehensive understanding and risk management, irrespective of whether the technology is developed in-house or obtained externally. As AI becomes ever more sophisticated, an important question is whether some third-party supply arrangements could become critical or important outsourcings, triggering specific FCA obligations.
So, while AI and ML hold the promise to revolutionise financial services, their full-scale, unchecked deployment might also result in unforeseen, possibly damaging, consequences. Therefore, the FCA report is a call to arms for the industry and regulators to grapple with AI's complexities and challenges to ensure that AI serves to benefit everyone in the financial ecosystem.
"The FCA clearly expects all regulated firms to ensure that its use of AI does not compromise a firm's ability to demonstrate compliance with its rules, including in relation to any services that are procured externally."
While the risks of harm to consumers are lower in the private markets than among firms serving the retail market, the FCA clearly expects all regulated firms to ensure that use of AI does not compromise the firm's ability to demonstrate compliance with its rules. The starting point is for firms is to think through – for the specific applications they have in mind – what the risks are, and which rules and regulations might be engaged. For example, using AI to process investor due diligence could raise anti-money laundering compliance issues, while GDPR and confidentiality issues are front of mind when personal information is made available to AI.
Good AI compliance in a firm will usually require a policy, with appropriate communications to staff. Many firms have already developed rudimentary policies, but these will need regular review and update. It is imperative that anyone handling commercially sensitive or confidential information understands how AI uses any data that it has access to – and who else inside and outside the organisation has access to it. As always, systems and controls are essential.
AI regulation is developing fast, not only in financial services but more generally, across Europe and North America – which means that firms will need to think about these issues wherever they operate – and so will their portfolio companies. As explored in our recent podcast with Cornelia Gomez at General Atlantic, responsible AI is a particular concern for investors in tech businesses. Those firms may be ahead of the curve in developing a robust approach to AI, but all firms will have to catch up.
This edition of Alternative Insights was written with assistance from TSBot, a Travers Smith AI tool. It was checked and edited by a human.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.