{"id":27815,"date":"2025-09-16T10:16:20","date_gmt":"2025-09-16T14:16:20","guid":{"rendered":"https:\/\/www.crim.ca\/?p=27815"},"modified":"2026-04-17T10:41:08","modified_gmt":"2026-04-17T14:41:08","slug":"evaluating-and-securing-banking-chatbots-technical-approaches-and-real-world-challenges","status":"publish","type":"post","link":"https:\/\/www.crim.ca\/fr\/evaluating-and-securing-banking-chatbots-technical-approaches-and-real-world-challenges\/","title":{"rendered":"Evaluating and securing banking chatbots: technical approaches and real-world challenges"},"content":{"rendered":"

Ce billet de blogue est \u00e9galement disponible en fran\u00e7ais<\/a> ci-dessous.<\/em><\/p>\n

In the banking sector, integrating chatbots can present major technical challenges, particularly in terms of performance measurement and risk management.<\/strong><\/p>\n

On Thursday, October 2 at noon<\/a>, Guillaume Delroeux, President of Prom\u00e9th\u00e9e Consultants<\/a>, and Marc Queudot, Practice Lead in Language Science and Technology at CRIM<\/a>, will share the results of a rigorous comparative study conducted on conversational assistants used by a range of major financial institutions in Canada.<\/p>\n

Performance indicators: granularity and relevance<\/h2>\n

The first step in evaluating an assistant is to define robust metrics.<\/p>\n

\u201cIn our analysis, we focused on simple answers. Is the answer to the user’s question correct?\u201d explains Marc Queudot in an interview in French.<\/p>\n

He distinguishes between metrics related to system performance and response accuracy, and those concerning user experience.<\/p>\n

\u201cIn well-designed assistants, the experience is fluid and the support is much higher\u2014you\u2019re offered the information you need rather than just an answer close to your question. It\u2019s subtle, but you can only really know by experiencing the bot.\u201d<\/p>\n

The technical analysis should also consider the assistant\u2019s ability to manage multi-turn conversations rather than isolated question-answer pairs, as well as conversational context\u2014two elements that complicate evaluation but are essential for advanced use cases.<\/p>\n

Risk management: intrusion testing and monitoring<\/h2>\n

Chatbots can come with multiple risks: hallucinations, bias, critical errors and damage to a brand’s image.<\/p>\n

Marc Queudot emphasizes the need for intrusion testing.<\/p>\n

\u201cThe classic approach in security for assistants is called red teaming, which involves a team trying to break the chatbot.\u201d<\/p>\n

This type of test helps identify vulnerabilities, especially those that could expose the organization to regulatory risks.<\/p>\n

He also highlights the importance of limiting transactional capabilities.<\/p>\n

\u201cOften, answering questions is enough for an assistant. You should think twice before adding features that increase risks.\u201d<\/p>\n

\"Marc
Marc Queudot, Practice Lead in Language Science and Technology at CRIM<\/figcaption><\/figure>\n

Ongoing improvement: automation and human oversight<\/h2>\n

Ongoing improvement relies on automated collection and analysis of conversations.<\/p>\n

\u201cAs usage grows, you\u2019ll gather data, detect conversations where performance is lacking and identify areas for improvement.\u201d<\/p>\n

Systems must group problematic cases to allow human experts to intervene on complex issues. This hybrid approach ensures rapid adaptation to new needs and effective correction of detected flaws.<\/p>\n

You should think twice before adding features that increase risks.<\/p><\/blockquote>\n

CRIM: expertise in risk mitigation<\/h2>\n

As a non-profit, CRIM offers a methodology for mitigating technological risks, combining experimentation, measurement and continuous optimisation.<\/p>\n

\u201cWhat we offer at CRIM is to carry out the technological risk mitigation and answer the question: can we develop a reliable, high-performing system and deploy it with minimal risk? That\u2019s our core expertise and it enables organiszations to make informed investment decisions.\u201d<\/p>\n

Webinar: results and recommendations<\/strong><\/h2>\n

On Thursday, October 2, at noon, the webinar will present the results of a comparative study on conversational assistants used by major Canadian financial institutions. Participants will benefit from the involvement of Prom\u00e9th\u00e9e’s team, which brings customer experience expertise to enhance the work’s impact.<\/p>\n

\u201cWe\u2019ll uncover the challenges faced by current systems. Then, we\u2019ll discuss how to evaluate these systems to ensure we have the right vision and path for improvement.\u201d<\/p>\n

On the agenda for the session, to secure and maximize the value of banking chatbots :<\/p>\n