AI agents are being deployed by consumers to navigate customer service calls, negotiate bills, and cancel subscriptions. While this offers a time-saving convenience for users, it presents challenges for human agents who must authenticate these AI callers and determine if they are authorized representatives or potential scammers.
The increasing use of AI agents in customer service interactions challenges traditional authentication and security protocols, potentially creating new avenues for fraud and impacting the efficiency and performance evaluation of human customer service representatives.
Consumers are increasingly deploying AI agents to navigate customer service interactions, leading to a new dynamic where human representatives are fielding calls from non-human entities. These AI agents are used to negotiate bills, cancel subscriptions, and gather information, often with the goal of giving companies a taste of their own medicine. While these AI callers are currently a small fraction of total call volume, their numbers are growing, prompting businesses to consider how to authenticate them, manage their interactions, and identify potential security risks.
Human agents report varied experiences, from AI agents that are progressively irritated when they lack answers to those that seem to be mapping out customer service systems. Some AI agents have an "attitude" and can even negatively impact human agents' performance reviews if they are not perceived as accommodating enough to the robot. Companies are grappling with policies on whether to engage with AI callers, how to verify their identity, and whether they are authorized to act on behalf of a customer. Security concerns arise as AI can be used by fraudsters to attempt to glean personal details.
Customer service outsourcing companies like TTEC have begun detecting these AI callers and are consulting with clients on how to handle them. While straightforward insurance preauthorization calls may work well, other interactions, such as those requiring personal information for verification, can be problematic for AI agents. The challenge lies in distinguishing between legitimate AI assistance, a customer convenience, and a potential security threat.
Pick the topics you care about. Get only what matters, on your cadence.