Table 1.

Known risks in conversational chatbots.

Risk category and description			Definition
Human rights
	Discrimination	The chatbot makes different recommendations or has a higher error rate based on the patient’s group (gender, ethnicity, race, religion, etc).
	Stereotyping	The chatbot interprets or uses language that propagates harmful prejudices, such as the inferiority of certain groups, sexualization, or lack of credibility.
	Exclusion	Development, governance, or use of the chatbot does not include certain already marginalized groups.
Data protection
	Lack of privacy	The data generated by the chatbot is not protected.
	Poor data governance	The data generated by the chatbot is governed improperly or without including the patient.
	Stigma	The data generated by the chatbot can lead to stereotyping or marginalizing certain individuals.
Technical
	Error tolerance	Errors, even if they are not discriminatory, cause harm to the patients.
	Overconfidence and trust decay	Patients place excessive trust in chatbots resulting in overconfidence and relative decay of trust in human health professionals.
	Technological solutionism	Investment in chatbot technology diverts from an actual societal problem.