Supervisory Authorities’ discussion paper on AI and machine learning - Which? response

Summary

Which? welcomes the opportunity to respond to the Supervisory Authorities’ discussion paper (DP) on artificial intelligence (AI) and machine learning. 

We are pleased that Supervisory Authorities are looking into how AI can be regulated, allowing for both innovation and robust consumer protection. Which? believes that innovation and consumer trust in AI will flourish if the right consumer protections are embedded into AI regulation. We welcome that the Supervisory Authorities are addressing this now to avoid any serious consumer harms emerging. 

We have identified priority concerns surrounding the use of AI in financial services:

• Biased or inaccurate data sets - leading to unfair or incorrect automated decisions on life-changing decisions for consumers.
• Unfair price personalisation - leading to consumer groups paying more for products and services without being aware, or excluding certain groups of consumers.
• Lack of transparency - threatening consumers' ability to identify and/or challenge unfair application of AI and undermining consumers’ trust in markets.
• Lack of cohesion in the current AI regulatory landscape - leading to businesses not applying the right protections to their use of AI. 

In response to these concerns, Which? recommends that the Supervisory Authorities:

• Demand that financial institutions are transparent about the data used in AI and that it is open to regular scrutiny to check for accuracy and bias.
• Ensure that businesses are transparent about where and how they use price personalisation algorithms.
• Consult the DRCF to ensure that AI best practice guidance for business is harmonised across the regulators.
• Ensure that the FCA are appropriately resourced with AI expertise and consult relevant bodies for advice.
• Ensure that consumer protections are enforced by the FCA where bad practice is found.