Current research reveals the negative effects of corporate social responsibility (CSR) on employees (e.g., CSR can lead to unethical behavior). In this study, we advanced the negative impact of CSR from reality to a hypothetical level, investigating whether counterfactual thinking that CSR could have been achieved but did not would lead to unethical employee behavior. In the seven studies, we designed a chatbot program based on the WeChat API and controlled different anthropomorphic levels to determine the chatbot's artificial intelligence (AI) level. Studies 1a and 1b show that counterfactual thinking regarding how companies could have engaged in CSR can induce unethical behavior, even when companies do not actually engage in CSR. Studies 2a and 2b indicate that unethical behavior can be induced by the counterfactual determinant of controllability. Studies 3a and 3b demonstrate that the degree of the controllability focus of counterfactual thinking and unethical behavior is positively correlated. Study 4 found that both counterfactual and prefactual thinking can induce unethical behavior. This study provides evidence on the impact of CSR, the function of counterfactual thinking, and the link between counterfactual thinking and the moral licensing effect.