AI agents are increasingly calling customer-service lines to negotiate bills, cancel subscriptions and handle routine tasks. They remain a small share of calls, but companies are weighing how to verify callers and limit fraud without adding hurdles for people.
A California insurance claims worker says the same AI agent, calling for an accident-law firm, reaches her a couple of times each shift; it cannot answer incident questions and can keep her waiting for an associate who never appears. At a major credit-card company, a retention agent says bots are rare and sometimes seem to map how to reach staff. Policy bars her from asking whether a caller is AI, and some account actions, including closing a card, must be done by the customer.
TTEC, a customer-service outsourcing firm, said it noticed AI callers about 18 months ago; clients generally asked staff to take calls cautiously. Agents can fail phone-number checks and customer challenge questions, while companies worry AI could let fraudsters place thousands of calls to gather personal data. Alorica's co-CEOs envision an AI-agent registration system to verify a bot's link to a person. A 2026 Accenture survey found 74% of more than 25,000 consumers across 16 countries would delegate routine tasks to AI, but 9% would let it make purchases independently.
