Is there observable behavior or evidence capable of distinguishing a liar from a truthful person? The question has intrigued for centuries, reflecting the inseparable nature of lying in our social interactions. Faced with the human inability to discern truth from falsehood and the limitations of polygraphs, the emergence of facial recognition technologies appears as a promising alternative. Although the use of facial expressions, including micro and macro-expressions, in lie detection is subject to debate, some researchers argue that these nonverbal cues could be revealing. This state-of-the-art review explores facial expression analysis through machine learning for lie detection, questioning the quality of data required to develop effective models.