The growing use of Artificial Intelligence (Al) in customer service has raised ethical concerns, particularly regarding the disclosure of Al identities, which may reduce customers’ purchase intentions and create a dilemma between ethical practices and profit maximization. While existing research has largely focused on Al's role in addressing customer needs and driving sales, the impact of Al identity disclosure during the customer acquisition phase - such as offering free product trials - remains underexplored. This study investigates the effects of Al identity disclosure during customer acquisition through a field experiment with 5,768 participants. The findings reveal that disclosing Al identities did not significantly affect customer’s willingness to accept the product. However, disclosure led to more concise communication, thereby enhancing sales efficiency. In terms of verbal behavior, disclosing the Al identity decreased customers' willingness to communicate, reduced politeness, and lowered emotional expression, with self-awareness serving as a mediator. When Al identity was not disclosed but the Al displayed high anthropomorphism, customers were more polite and communicative, although their product acceptance did not increase. These results contribute to the understanding of consumer - Al interactions and provide actionable insights for businesses aiming to optimize AI use in customer service.