 By Tim Sullivan, Asset Management Industry Leader, US and Richard Langdon Asset Management Industry Leader, GB from Willis Towers Watson
If you’ve attended an insurance conference, looked at a newsfeed, or scrolled through LinkedIn, you’ve likely encountered AI-related content.
Be that as it may, there is good reason for this topic to be at the forefront of insurance discussions. The benefits of AI, particularly generative AI, are seemingly endless, but the associated risks are complex and evolving, and their impact on an organization’s risk profile cannot be taken lightly. As a risk transfer mechanism, insurance should be incorporated into AI discussions to determine what impact, if any, these risks may pose to an organization’s insurance portfolio.
The following commentary analyzes AI through the lens of insurance, specifically highlighting key issues privately held asset managers should consider when reviewing their financial and executive liability insurance programs.
AI and the asset management industry
Judging by the recent hype, one would think AI was invented last week. However, trading systems, robo-advisors, legal and compliance, research and analysis, marketing, and customer service are just a few areas where AI has already been incorporated, to varying degrees, within the asset management industry. What has changed, and what will continue to evolve, is the advancement and sophistication of this technology and its wider proliferation across all industries, including asset management.
To remain competitive, it is expected that asset managers will continue to incorporate AI even deeper within their operations. Further, AI’s potential to create greater value for investors will most certainly become an avenue for differentiation within the industry. With these advancements, however, comes increased risks, including regulatory scrutiny over how AI is being utilized and deployed, and what statements are being made to investors regarding its benefits.
Evolving Regulatory Landscape
Within the US, regulatory focus has noticeably increased over the past year. For example, in July 2023, the SEC released a proposal to address conflicts of interest where analytics, including AI, are used to make investment predictions. Then in December 2023, the Wall Street Journal reported that the SEC was conducting sweeps of advisers, seeking information on marketing, algorithmic models, training, compliance and oversight. Further, the White House issued an Executive Order outlining AI best practices, while certain US states and European countries enacted and/or proposed laws or regulatory frameworks focused on these risks. As AI continues to evolve, regulators will likely take additional steps to protect investors, especially those at the retail level.
The UK and the European Union look set to adopt different approaches to the regulatory oversight of AI. The European Parliament have recently approved the EU AI Act which will seek to introduce a comprehensive regulatory framework for the use of AI which will rate the risks associated with AI platforms, imposing strict requirements on those deemed to carry higher risks. Firms generally will be required to adhere to higher standards of data governance and risk management as the EU looks to set the standard for future AI regulation.
The UK Government, however, have favored a less prescriptive, outcome-based approach which is underpinned by five core principles: safety; security and robustness; appropriate transparency and explainability; fairness; accountability and governance; and contestability and redress. Their aim is to strike a balance between regulatory oversight and promoting innovation, however asset managers will still have to adapt their governance frameworks to account for the nuances of AI and machine learning technology, especially with respect to individual accountability under the Senior Manager and Certification Regime (SMCR).
AI washing
Given all this attention, it was not much of a surprise when the SEC announced in March 2024 that it brought charges against two advisers alleging several AI-related violations, including “AI washing”. This latest use of “washing” generally refers to the alleged false and/or misleading statements regarding the use and value of AI within the investment management process. While the $400,000 penalties imposed upon the advisers may seem relatively tame, the SEC’s actions in this matter provided useful insight into how the regulators might approach this issue more broadly moving forward, and what advisers should consider before promoting their use of AI.
Asset management risks
While the benefits of AI are clear, there are numerous risks that must be taken into consideration. Though these will most certainly evolve, examples of AI risks applicable to asset managers include, but are not limited to, the following:
Data Accuracy: Incorrect AI inputs leading to erroneous outputs may result in poor investment decisions and losses. Such losses may lead to investor and regulatory claims under management and professional liability insurance policies, alleging a failure to provide proper oversight by the directors and officers, as well as errors and omissions in the performance of, or failure to perform, investment management services.
Job Displacement: Workforce efficiency may lead to reductions-in-force which, if disproportionately impacting certain classes of employees, may trigger discrimination claims under employment practices liability (EPL) insurance policies.
Biased Data: Bias and discrimination are the top risks today. Biased outputs resulting from biased inputs (unintentional or otherwise) may lead to discriminatory activity, such as when AI-driven hiring tools inadvertently decline employment candidates over a certain age, possibly triggering EPL claims activity.
Regulatory Actions: New or modified regulatory frameworks increase the risk of formal and informal regulatory investigations, enforcement actions, and litigation, potentially triggering coverage under management and professional liability policies.
Cybersecurity: The misuse of AI tools to execute fraudulent attacks and compromise data raises a host of concerns that may lead to claims under a cybersecurity and/or crime policy.
Application Failure: The failure of AI applications to perform as intended may generate a claim triggering any applicable technology or media liability insurance coverage.
Trade Execution: Erroneous trades made via AI-supported trading systems may result in investor damages, potentially leading to claims under the “cost of corrections” section of an asset managers’ professional liability policy.
Benefit Plans: Should funds that are included as investment options within sponsored benefit plans suffer AI-related losses, plan sponsors and their fiduciaries may be the target of claims alleging breach of fiduciary duty, which may trigger coverage under the sponsor’s fiduciary liability (also known as pension trust liability) insurance program.
Intellectual Property: When building out proprietary AI models, algorithms, and other intangible assets using 1st or 3rd party data, asset managers must ensure a thorough vetting process to mitigate the risk of trade secret misappropriation and copyright infringement. Litigation in this area is rapidly evolving and coverage for such claims is typically sought under an IP liability insurance program.
As with all scenarios, the applicability of the referenced insurance programs will depend on the specifics of the particular claim and the language within the actual insurance policies.
Key insurance renewal considerations
While each insurer will have their own perspective, and each insureds’ risk profile is unique, asset managers should be prepared to address AI-related questions as part of their financial and executive liability insurance renewals. Areas of interest may include, but are not limited to, the following:
Overall understanding of the potential risks presented by AI;
Types of AI currently used and/or expected to be used in the near future;
Those internal functions currently utilizing AI and to what degree;
Credentials of those overseeing and monitoring the use of AI;
The (human) controls monitoring and reviewing the use and/or output of AI;
Extent to which AI training for employees is conducted;
Process for vetting public statements made about the benefits of AI to investors;
If and when AI may replace roles currently held by employees;
Process for assessing evolving third-party risk from AI vendors and other third parties;
Extent of indemnification received from AI vendors and other relevant parties;
Extent of indemnification provided to clients; and
Changes to cybersecurity, fraud, or other risk mitigation frameworks in response to AI.
Conclusion
The opportunities and benefits of AI are extensive, but it is critical that firms have a close eye on the risks and the evolving regulatory and legal landscapes applicable to AI. During the renewal underwriting process, it is important for asset managers to demonstrate their understanding of AI risks and how they may affect both their firm and their clients. More importantly, it is critical to clearly communicate the policies and procedures in place to mitigate those risks. In doing so, asset managers are likely to achieve a more favorable outcome when their insurance programs renew.
|
