It has been clear for some time that the increased prevalence of AI would not automatically lead the FCA to introduce a whole raft of new regulations around it…and that is the right approach. However, with AI a constant presence in the media and our lives in general, it is clear that senior management at firms will often feel the pressure or the need to adopt some form of AI, to prove that they are keeping up with the pack.
However, whilst the FCA recognise the transformative potential of AI in financial services, their publication on AI in financial services reinforces an important point: adopting AI is not a box to tick. Firms adopting AI must also consider the fundamentals; regulatory responsibilities, governance arrangements and the potential impact on customer outcomes.
For firms considering introducing AI, here are eight questions worth asking before taking the next step.
Consumer Duty
1. Customer outcomes: How could the proposed use of AI affect customer outcomes, both positively and negatively? Have the potential risks been properly assessed?
2. Vulnerable customers: Could the system disadvantage customers in vulnerable circumstances or make it harder for them to access appropriate support?
3. Consumer understanding: If AI is used to communicate with customers or provide information, how will you assess whether its outputs are clear, accurate and support informed decisions?
4. Testing: How will you test the system before implementation to identify potential bias, inappropriate outputs or other risks to customers? Will testing reflect the different circumstances of your customer base?
Governance and Oversight
5. Accountability: Who will be responsible for the system, its risks and the outcomes it produces? Are roles and responsibilities clear before implementation?
6. Risk assessment and approval: How will AI risks be assessed, challenged and approved? Are the right people involved before the system goes live?
7. Management information: What MI will you need to monitor the system's performance and its impact on customers? How will you identify unexpected outcomes or emerging risks?
8. Ongoing oversight: What arrangements will be in place to review performance, investigate issues and take corrective action? How will you ensure human oversight is meaningful rather than simply a sign-off exercise?
These questions are not about introducing another layer of bureaucracy. They are about considering the implications before implementation, establishing appropriate controls and being able to demonstrate that those controls remain effective.
The key question is not simply whether AI can make a process more efficient, but whether it can be introduced in a way that supports good customer outcomes and effective governance.
At Danbury Compliance, we support firms in reviewing and strengthening their compliance and governance arrangements through services including:
· Consumer Duty Governance Review – examining governance arrangements and MI to help assess whether firms can demonstrate effective oversight of Consumer Duty and customer outcomes.
· Board Effectiveness Review – available as part of our ongoing retainer support, helping firms assess the effectiveness of board-level governance, challenge and oversight.
Both can help firms take a closer look at their existing arrangements, identify areas for improvement and strengthen their approach to effective governance. With those key foundations in place, this can help firms create the conditions for a more successful adoption of AI.
Find out more:
🔗 Consumer Duty Governance Review: https://danburycompliance.co.uk/consumer-duty-governance-review.php
🌐 Danbury Compliance: https://danburycompliance.co.uk/