With the growing use of artificial intelligence (AI) in financial services, we were concerned about data governance, fairness, algorithm biases and accountability of AI tools implemented by Royal Bank of Canada.
Background
Royal Bank of Canada (RBC) engages in the provision of banking and financial solutions. It operates through the following segments: Personal and Commercial Banking, Wealth Management, Insurance, Investor and Treasury Services, Capital Markets, and Corporate Support.
Across these segments, RBC offers a broad suite of financial products and services in Canada with operations in 29 countries and therefore plays a pivotal role in establishing responsible policies in the use of AI in advancing its market position.
Our engagement
With the growing use of artificial intelligence (AI) in the financial services sector, we were concerned about data governance, fairness, algorithm biases and accountability during the implementation of AI tools at RBC. We believe that companies should publish governance principles for the ethical use of AI in a format accessible to stakeholders.
We were concerned about data governance, fairness, algorithm biases and accountability during the implementation of AI tools at RBC.
Accordingly, as the bank’s implementation of AI expanded, we engaged to understand the principles governing the use of AI in a way that explains the ethical considerations of its implementation and ensures transparency, accountability, and fairness to users.
In Q3 2020, we began conversations regarding AI and asked the company to consider publishing its AI principles and assigning board-level responsibility to a director to demonstrate good governance and hedge for emerging risks.
In early 2021, we met with the company to discuss the strategy, utilisation and governance of AI. We were pleased to learn that the AI topic had recently been presented to the bank’s board of directors and that while the board does not have a designated AI expert, the topic is discussed at board meetings.
We were pleased to learn that the AI topic had recently been presented to the bank’s board of directors and that while the board does not have a designated AI expert, the topic is discussed at board meetings.
As the bank’s AI-related offerings continue to evolve, we encouraged it to map its AI footprint as a means for strengthened governance and risk management oversight. The team said it has been working towards creating an AI risk profile to share with the board. At RBC, AI augments customer tools and does not perform decision-making processes without appropriate human oversight.
The company had not published its principles of ethical AI on the RBC website, but has published it on the Borealis AI site and had an internal white paper highlighting such principles, and emphasised that its practices align with the Borealis AI institute, a research centre created by the bank driving research on responsible AI.
The company agreed to consider our request to disclose its principles publicly, as we emphasised the value of doing this to alleviate stakeholder concerns.
The company agreed to consider our request to disclose its principles publicly, as we emphasised the value of doing this to alleviate stakeholder concerns. It currently meets regulatory guidance for model risk management which includes validation of AI. The company welcomed receiving our artificial intelligence materials.
In Q1 2022, during a meeting to discuss the bank’s 2022 proxy statement, the company noted that the board’s 2021 education agenda included AI risk considerations and oversight. We continued to press the company for more clarity on board and senior management level oversight of AI.
Changes at the company
In 2023, the company received a shareholder proposal on AI governance and enhancing the mandates of the governance committee and the risk management committee to include an ethics component regarding the use of artificial intelligence. The proposal was withdrawn given that the
company had met the asks of the proposal but had not explicitly disclosed the principles.
In response to the proposal, the company disclosed its five principles for responsible use of AI, along with the board’s oversight on strategic direction, plans and priorities of the bank and how it ensures these facets align with the risk appetite framework which the board approves annually.
The response to the withdrawn shareholder proposal noted that the board receives regular reports from the business segment heads, including on initiatives involving the use of AI and the risk committee reviews significant and emerging risks, including those related to the adoption and use of AI.
In response to the proposal, the company disclosed its five principles for responsible use of AI, along with the board’s oversight on strategic direction and priorities.
Further, in June 2023, we met with the global head of market and counterparty credit risk and the interim head of Borealis AI, RBC’s lab for AI. The company stated that no single board member has AI oversight, but it relies on the expertise of all directors, including those with technology/ digital competencies.
Together, with proxy disclosures and engagement context, we believe the company has sufficiently addressed board-level AI oversight.
Together, with proxy disclosures and engagement context, we believe the company has sufficiently addressed board level AI oversight.
Later, we wrote to the company and applauded Borealis AI’s leadership in responsible AI and its work on the RESPECT AI pillars which identify robustness, fairness, model governance, data privacy and explainability as critical for responsible and safe AI. We acknowledged the company’s disclosure of its five principles for the responsible use of AI and the board’s oversight on AI.
Taken together, the RESPECT AI pillars and the disclosure on RBC’s principles and oversight are considered emerging best practice in disclosing ethical AI principles. As such, we deem our ask substantially met.