The field of legal AI is rapidly evolving, with a focus on developing more sophisticated and transparent models for legal reasoning and decision-making. Recent research has highlighted the importance of addressing issues such as motivated reasoning, hallucination, and uncertainty in legal question answering. To this end, innovative approaches such as multi-agent workflows, structured prompting methodologies, and retrieval-augmented generation have been proposed. These methods aim to improve the accuracy, reliability, and explainability of legal AI systems, and have shown promising results in various benchmarks and evaluations. Notably, the integration of domain-specific knowledge bases and the use of large language models have enabled the simulation of judicial deliberation dynamics and the generation of legally sound instructions and arguments. Overall, the field is moving towards more comprehensive and nuanced models of legal reasoning, with a emphasis on transparency, accountability, and scalability. Noteworthy papers include: Modeling Motivated Reasoning in Law, which investigates how LLMs respond to prompts conditioned on different legal roles. L-MARS, which proposes a multi-agent workflow with orchestrated reasoning and agentic search to reduce hallucination and uncertainty in legal question answering. KoBLEX, which introduces a benchmark for open-ended and provision-grounded legal question answering and proposes a method for parametric provision-guided selection retrieval.