How the EU AI Act will reshape customer experience design
Customer experience (CX) leaders working in Europe and beyond risk millions in fines starting next week as the EU AI Act’s transparency and enforcement rules kick in.
Though originally introduced in 2024, the latest obligations under Article 50 include disclosing AI use to customers upon first contact, having a practical way to connect with a human being and clearly communicating that emotion recognition systems have been deployed. Firms must also be transparent about the use of deepfakes and AI-generated text on public-interest matters.
Non-compliance with the EU AI Act can attract fines of up to EUR 15 million, or three per cent of a firm’s worldwide annual turnover. The latest obligations will make CX processes one of the essential ways regulators ensure AI use is properly governed.
Speaking in a live broadcast about the issue on LinkedIn last week, experts with Austin, Tex.-based CX technology provider TTEC Digital said compliance could be challenging, given how AI is increasingly embedded in a wide variety of platforms and application integrations.
“It’s not only about the chatbot speaking to a customer,” said Georgi Georgiev, the firm’s senior corporate counsel for the EMEA region. “It can include various other use cases, like for example, agent assist tools, or quality monitoring, or post interaction insights. There are a lot of areas across the CX process where AI exists and operates, but it’s not immediately visible.”
The EU AI Act put the onus on organizations to identify where AI is being used across their technology stack, but CX leaders aren’t always deliberately buying and deploying it, said Amanda Pank, TTEC Digital’s senior manager of strategy and operations. Instead, AI might come through an upgrade to a current platform or deep within QA tools.
“Make a list or an inventory of what you’ve got, because until you know that, it’s really difficult to make that assessment of the risk and the impact,” she said. “Then it’s a matter of mapping out those customer journeys because it’s really key to see that end to end piece of where AI is appearing and how it’s working.”
While generative AI tools may be summarizing contact center interactions following a customer call, for example, agentic AI tools might be making decisions on behalf of a customer, an employee or both.
The urgency to comply isn’t limited to CX leaders operating solely within Europe. Much like GDPR, a law focused on privacy protection, the AI Act will apply to any brand who sells or provides services to those living within the EU. Brands in North America who comply with Article 50 now may be in a better position as regulators in other jurisdictions develop legislation of their own, Georgiev suggested.
“I expect that the EU AI Act will become the practical benchmark for AI-related work across the globe, so its relevance and applicability will only increase,” he said.
Pank said responsible usage really means ensuring customers never feel deceived or confused about where AI is being woven into an experience.
“When working with clients, listening to interactions from customers today, it’s not unusual to hear a customer say, ‘Am I speaking to a person or are you a robot?’” she said. “And I think being really clear in setting that expectation, so the customer understands what they’re interacting with, gives them that chance to make an informed choice about how they’re interacting, because obviously it’s going to involve their data.”
Pank also she’s been getting a lot of questions from customers about who “owns” AI transparency, but suggested it’s a responsibility that should be well understand and acted upon throughout an organization.
While it regulation is often seen as a barrier to innovation, Georgiev said the EU AI Act is really more of a framework for deploying the technolgy in a way that everyone – including clients, regulators, employees and customers – can trust. It could also help businesses grow.
“The commercial opportunity is to move from AI experimentation to AI adoption at scale, but in a responsible and compliant manner, which, at the end of the day, I believe will differentiate successful businesses from those who use AI without the proper command framework to support it,” he said.
