Companies that want to do business in the European Union will now have to disclose when customers interact with AI and when content is AI generated.
On Aug. 2, the EU begins enforcing key transparency regulations of the EU AI Act. While some aspects of the EU AI Act were delayed, Article 50 requires businesses inform people they are interacting with AI, mark AI-generated synthetic content, disclose emotion recognition and biometric categorisation, and label deepfakes and AI-generated public interest text.
The date also marks the creation of the authority in charge of that enforcement, called the EU AI Office.
The EU AI Act regulations are an opportunity to build trust with consumers through transparency and revise AI governance frameworks, experts told CX Dive.
“The organization and CX leaders shouldn't see this as another regulatory requirement,” Mahadevan Meikum Perumal, VP and head of delivery and client success of EMEA at TTEC Digital, told CX Dive. “That's the wrong way of looking at it. To me, it's an opportunity.”
The EU AI Act provides consumers with what they clearly want: transparency.
“With these transparency requirements, what I keep telling the companies I talk to is, ‘Forget about the regulators. In the last 24 months, consumers' data across geographies is very clear. Consumers want to know when they are exposed to AI,’” Enza Iannopollo, principal analyst at Forrester, told CX Dive. “We are talking about over 70% of consumers in almost every given market that we survey.”
Establishing AI systems with customers in mind
The regulations are also an opportunity to establish AI governance with the customer in mind, Perumal said. That begins with taking inventory of what’s already deployed.
“What's important is to have the inventory of your systems first and foremost for the second of August deadline,” Perumal told CX Dive. “Many organizations don't realize the amount of AI that is embedded within the organizations.”
Perumal encourages CX leaders to work in tandem with their CIOs, CTOs and chief data officers — and plan to keep in touch.
“The wrong thing to do is to limit it to a legal or an IT problem,” Perumal said. “The common thing that we see is [making] it a CX versus a legal thing. If it's anything, it is an organization-wide initiative.”
As companies deploy more AI applications, complying to the EU AI Act regulations will be an ongoing project.
“So many times the AI innovation is not going to be contained; it's going to be evolved,” Perumal said. “Making the system compliant is the beginning of the journey, not the end of the journey. It is about establishing the governance and constantly revisiting the governance.”
What consumers can expect
When EU consumers interact with an AI chatbot or other AI application on or after Aug. 2, they can expect to see more communications or attempts to convey that they are interacting with AI.
As with the EU’s General Data Protection Regulation legislation, how a company discloses is just as much of an experience hurdle as it is a legal and technical challenge. Clear, unobtrusive disclosure can generate trust. Unhelpful or obtrusive disclosure creates confusion.
Not all of businesses' early attempts for transparency will be frictionless.
“I expect to see some of these implementations to be off in terms of experience,” Iannopollo said. “You might encounter more banners or more elements that are a little bit intrusive compared to the experience that you were used to, but I think businesses will get a little better in then refining some of the strategies around how they communicate around these things.”
Eventually, Iannopollo expects businesses to display more symbols indicating AI.
Building trust through responsible governance
Forrester research has consumers trust AI more when companies are transparent in how they're deploying it.
“Transparency is one step into making people feel more comfortable, helping them become more familiar, and helping them understand the trade-offs, which I really think is a fundamental part here,” Iannopollo said.
A Genesys survey from last year found that 4 in 5 consumers want businesses to deploy AI with robust guardrails because they don’t trust the technology. But trust increases when consumers know that a human is available or in the loop. Good AI governance also builds trust as successful AI deployments are more likely to be used again.
Once CX leaders address regulatory questions, they should use the regulations as an opportunity to get to know their customers, Iannopollo said.
“What I really would do is look at the customer base that I serve and make sure that I start understanding how is this consumer population feeling about the way I'm showing up with this AI? Is that changing the experience for them for better or worse?”
