The field of large language models (LLMs) is rapidly evolving, with a growing focus on their applications in education and ethics. Recent developments have highlighted the need for more nuanced and context-aware approaches to evaluating LLMs, particularly in areas such as privacy, moral reasoning, and speciesism. Researchers are exploring new frameworks and benchmarks for assessing LLMs' capabilities and limitations, including their ability to function as artificial moral assistants and their potential for reproducing entrenched cultural norms. Notable papers in this area include SproutBench, which introduces a benchmark for evaluating LLMs' safety and ethics in applications targeting children and adolescents, and Beyond Ethical Alignment, which evaluates LLMs' moral capabilities and highlights the need for dedicated strategies to enhance moral reasoning capabilities. Overall, the field is moving towards a more comprehensive understanding of LLMs' potential and limitations, and the development of more effective and responsible approaches to their deployment in education and other areas.