The Risks of Using Chatbots in the Financial Industry and How to Foster Consumer Confidence

Print PDF


Amundsen Davis Banking & Finance Alert

Recently, the Consumer Financial Protection Bureau (CFPB) released a report on the use of chatbots, such as ChatGPT, etc., by financial institutions. The report highlighted that chatbot use, as automated teller machines and offshore customer call centers before them, are increasingly used by financial institutions to streamline processes and lower operating costs. The CFPB report details a number of challenges in the use of chatbots, as well as guidance for successful deployment by financial institutions.

As chatbots proliferation rises, the CFPB is fielding an increasing number of consumer complaints related to chatbots of various sophistication. Among the most frequent complaints are:

  • Limited Ability to Solve Complex Problems: This is particularly true if the chatbot is labeled as “artificial intelligence,” while its capabilities are limited to more generic customer information functions.
  • Difficulties in Recognizing Customer Disputes: Chatbots frequently cannot identify the nature of customer disputes or direct the consumer to the appropriate resource as efficiently as human customer service representatives.
  • Inaccurate Datasets: Even the most popular chatbots cannot always determine when a dataset it is pulling from is inaccurate, which can lead to inaccurate responses.
  • “Doom Loops”: Chatbots may automate a response to serve continuous, unhelpful, and circuitous jargon without an ability to offramp to a human customer service representative.
  • Lack of Problem-Solving Optimization: If a customer problem falls outside a chatbot’s capabilities to assist, problems can rapidly escalate and require many more resources to resolve.
  • Technology-Specific Access Issues: Chatbots’ system downtime or a customer’s lack of technology compatibility can hamper customer access to immediate communication with their financial institution.
  • Security Risks: Chatbots can be susceptible to data hacking or phishing by hostile chatbots.
  • Secure Personal Identification: Because customers may input personally identifiable information into chatbots, chatbot logs should be secured using the same protocol that protects financial institution customer’s personally identifiable information.

In response to the CFPB’s report, financial institutions would be well-advised to ensure chatbots are only used for appropriate, informational tasks. They should be designed to ensure access across the board of technology capabilities of their customers. Chatbots should have offramps to human customer service reps that are more knowledgeable, and these offramps should be easily identified by consumers. Finally, chatbots that collect any personally identifiable information should have its chatlogs secured in accordance with financial institutions personal information protocols.

Any new technology which has the ability to streamline processes and potentially lower costs is one that cannot be ignored. Per the CFPB’s guidance, financial institutions should continue to integrate it in accordance with longstanding financial customer service principles: prioritize consumer access, results, and trust and confidence in their financial institutions, regularly audit chatbot usage to ensure it is not deployed beyond its capabilities, and ensure chatbots and chatbot logs are appropriately secured from outside intrusion. In doing so, financial institutions can ensure that they remain legally compliant and business efficient.

Jump to Page

This website uses cookies to improve functionality and performance. If you choose to continue browsing this website, you consent to the use of cookies. Click here to read about our privacy and cookie policy.