Ethical LLMs 2025: The first Workshop on Ethical Concerns in Training, Evaluating and Deploying Large Language Modelshttps://sites.google.com/view/ethical-llms-2025 @ RANLP2025https://ranlp.org/ranlp2025/
Call for papers:
Scope Large Language Models (LLMs) represent a transformative leap in Artificial Intelligence (AI), delivering remarkable language-processing capabilities that are reshaping how we interact with technology in our daily lives. With their ability to perform tasks such as summarisation, translation, classification, and text generation, LLMs have demonstrated unparalleled versatility and power. Drawing from vast and diverse knowledge bases, these models hold the potential to revolutionise a wide range of fields, including education, media, law, psychology, and beyond. From assisting educators in creating personalised learning experiences to enabling legal professionals to draft documents or supporting mental health practitioners with preliminary assessments, the applications of LLMs are both expansive and profound. However, alongside their impressive strengths, LLMs also face significant limitations that raise critical ethical questions. Unlike humans, these models lack essential qualities such as emotional intelligence, contextual empathy, and nuanced ethical reasoning. While they can generate coherent and contextually relevant responses, they do not possess the ability to fully understand the emotional or moral implications of their outputs. This gap becomes particularly concerning when LLMs are deployed in sensitive domains where human values, cultural nuances, and ethical considerations are paramount. For example, biases embedded in training data can lead to unfair or discriminatory outcomes, while the absence of ethical reasoning may result in outputs that inadvertently harm individuals or communities. These limitations highlight the urgent need for robust research in Natural Language Processing (NLP) to address the ethical dimensions of LLMs. Advancements in NLP research are crucial for developing methods to detect and mitigate biases, enhance transparency in model decision-making, and incorporate ethical frameworks that align with human values. By prioritising ethics in NLP research, we can better understand the societal implications of LLMs and ensure their development and deployment are guided by principles of fairness, accountability, and respect for human dignity. This workshop will dive into these pressing issues, fostering a collaborative effort to shape the future of LLMs as tools that not only excel in technical performance but also uphold the highest ethical standards.
Submission Guidelines We follow the RANLP 2025 standards for submission format and guidelines. EthicalLLMs 2025 invites the submission of long papers, up to eight pages in length, and short papers, up to six pages in length. These page limits only apply to the main body of the paper. At the end of the paper (after the conclusions but before the references) papers need to include a mandatory section discussing the limitations of the work and, optionally, a section discussing ethical considerations. Papers can include unlimited pages of references and an unlimited appendix. To prepare your submission, please make sure to use the RANLP 2025 style files available here:
* Latexhttps://ranlp.org/ranlp2025/wp-content/uploads/2025/05/ranlp2025-LaTeX.zip * Wordhttps://ranlp.org/ranlp2025/wp-content/uploads/2025/05/ranlp2025-word.docx
Papers should be submitted through Softconf/START using the following link: https://softconf.com/ranlp25/EthicalLLMs2025/ Topics of interest The workshop invites submissions on a broad range of topics related to the ethical development and evaluation of LLMs, including but not limited to the following.
1. Bias Detection and Mitigation in LLMs Research focused on identifying, measuring, and reducing social, cultural, and algorithmic biases in large language models.
2. Ethical Frameworks for LLM Deployment Approaches to integrating ethical principles—such as fairness, accountability, and transparency—into the development and use of LLMs.
3. LLMs in Sensitive Domains: Risks and Safeguards Case studies or methodologies for deploying LLMs in high-stakes fields such as healthcare, law, and education, with an emphasis on ethical implications.
4. Explainability and Transparency in LLM Decision-Making Techniques and tools for improving the interpretability of LLM outputs and understanding model reasoning.
5. Cultural and Contextual Understanding in NLP Systems Strategies for enhancing LLMs’ sensitivity to cultural, linguistic, and social nuances in global and multilingual contexts.
6. Human-in-the-Loop Approaches for Ethical Oversight Collaborative models that involve human expertise in guiding, correcting, or auditing LLM behaviour to ensure responsible use.
7. Mental Health and Emotional AI: Limits of LLM Empathy Discussions on the role of LLMs in mental health support, highlighting the boundary between assistive technology and the need for human empathy.
Organisers
Damith Premasiri – Lancaster University, UK Tharindu Ranasinghe – Lancaster University, UK Hansi Hettiarachchi – Lancaster University, UK Contact If you have any questions regarding the workshop, please contact Damith: d.dolamullage@lancaster.ac.uk