Trust and Ethics in Chatbots: A Framework for Responsible AI Interactions

The increasing deployment of chatbots across sectors such as education, healthcare, and customer service has raised the need of user interactions with chatbot be safe, fair, responsible and respectful. These expectations from a chatbot could be collectively named as ethical behaviour or ethical concerns in AI applications. Human beings by default are expected to possess moral values, ethics and fairness. Indeed, critical AI applications should also be aligned with these qualities as they deal with real time scenarios and people. Ethical concerns include human agency and oversight, technical robustness and safety, transparency, non-discrimination and fairness, accountability, social well-being, privacy and data governance. This paper explores the ethical implications of chatbots that exhibit varying responses due to fluctuations in dataset intents and patterns. Inconsistent behaviour can undermine user trust, introduce bias, and compromise decision-making, especially when the chatbot is used in sensitive domains. The study investigates how changes in the structure, labelling, or balance of intent patterns influence chatbot performance, potentially leading to unequal treatment of users or miscommunication on a university chatbot. Ethical principles in context to correctly classifying intent proves chatbot is technically robust with accuracy rate of 98% and a safety score of 87.65%, proves diversity and fairness with a fairness score of 84.5%, and existing frame works IBM AIF Fairness 360 and Microsoft RAI Toolkit is used for comparison with accuracy of 99.5 and 99.12% respectively and data governance achieved with a data quality score of 96.8% and contributes Societal and environmental well being with a sustainability score of 84% . Recommendations are proposed for implementing auditability, explainability, and robust validation processes to ensure that chatbot systems remain ethically compliant even under dynamic or evolving datasets. The research underscores the need for continuous monitoring and ethical auditing mechanisms to safeguard against unintended harms and promote responsible AI deployment.To add on value two additional domains healthcare and customer service chatbot are used for comparison.