This study aimed at assessing the use of artificial intelligence in academic writing by identifying ethical and unethical practices in the use of the technology. Ethics in academic writing has to do with principles and practices which academics have to apply and use to promote honesty and integrity in preparation and dissemination of their scholarly works. The study used qualitative research that uses in-depth interviews to collect data from academicians working in three higher learning institutions in Tanzania. A total of 17 academicians were sampled with the use of purposive sampling technique. A sample size of 17 was determined by the saturation principle. Thematic analysis was used to analyze collected data. The study identified seven ethical practices which are brainstorming research ideas, identifying relevant models/frameworks/theories, reading concepts generated by AI, editing and proofreading, data analysis and translation. Three unethical practices were identified which include not acknowledging the use of AI, overreliance on AI and using AI to fabricate data. The line between ethical and unethical practices is drawn by considering the extent to which AI assisted a researcher in writing academic documents. This extent may be determined by examining whether what AI did in a particular academic work could have been done by asking for assistance from a natural person and whether what AI did could have been done by any other technology which is acceptable in academic writing. The study recommends various measures to be undertaken to avoid malpractices in academic writing.