As artificial intelligence is rapidly developing and becoming a growing presence in healthcare communication, a new study addresses a concern that large language models (LLMs) can reinforce harmful stereotypes by using stigmatizing language.