Integrating Large Language Models (LLM) like ChatGPT in marketing practices is transforming the field because of the ability of such technologies to allow for personalized experiences and enhanced brand engagement. Apart from the emerging possibilities, some concerns are being raised, one of them being AI hallucinations (AIH), present in the results of such models, which can present risks to marketing outcomes, including damaging brand image and consumer trust. This study aims to shed light on AIH's impact on content accuracy and consumer behavior by performing a systematic literature review. The findings highlight increased customer complaints and harm to brand perception. To mitigate these risks, strategies like human-in-the-loop (HITL) systems and retrieval-augmented generation (RAG) models are recommended, with a call for further research to improve LLM robustness and protect the integrity of AI-generated content.