Loneliness, characterized by a perceived gap between desired and actual relationships, is a growing concern in the digital age and is particularly pronounced in individuals with autism due to social communication challenges and feelings of social isolation. This paper investigates the potential of empathic social chatbots to mitigate loneliness in autistic individuals, addressing challenges stemming from the distinct emotional expressions and non-verbal communication patterns of autistic individuals, which often differ from neurotypical norms that these algorithms are trained on. These discrepancies underscore the necessity for more inclusive and adaptive empathic AI technologies to better cater to the diverse emotional needs of autistic users. Addressing these challenges is crucial for developing effective tools to reduce loneliness for autistic individuals, highlighting the importance of ethical considerations in the advancement of empathic AI technologies.