Guest Editor(s)
Prof. Leonardo Ranaldi
Email: leonardo.ranaldi@uniroma2.it
Affiliation: Human-Centric ART, University of Rome Tor Vergata, Rome, Italy; School of Informatics, University of Edinburgh, Edinburgh, United Kingdom
Homepage:
Research Interests: LLM, reasoning AI
Dr. Pucci Giulia
Email: giulia.pucci@abdn.ac.uk
Affiliation: Department of Computing Science, University of Aberdeen, Aberdeen, United Kingdom
Homepage:
Research Interests: safety LLMs
Summary
Recent advances in artificial intelligence have produced models with remarkable capabilities in language, vision, and decision-making. However, their apparent proficiency often masks a reliance on shallow heuristics, pattern matching, or socially driven agreement rather than robust and reliable reasoning. This limitation poses significant challenges for AI safety, particularly in high-stakes settings where errors, misalignment, or unverified inference can lead to harmful outcomes. As a result, understanding and strengthening the reasoning processes underlying AI systems has become a central concern for trustworthy and accountable deployment.
This Special Issue focuses on Reasoning for Safety in Artificial Intelligence, bringing together research that explores how explicit, structured, and verifiable reasoning mechanisms can enhance safety, robustness, and oversight. It covers a broad range of approaches, including optimisation-based reasoning, agentic and multi-agent systems, debate and self-critique frameworks, symbolic and dialectical methods, and reasoning under uncertainty. Particular attention is given to the evaluation and interpretability of reasoning processes, as well as to mitigating pseudo-reasoning and sycophantic behaviours in large language and multimodal models.
By emphasizing methodological and technical advances in reasoning-aware AI safety, this Special Issue aims to promote principled frameworks for trustworthy, interpretable, and reliable intelligent systems.
Keywords
artificial intelligence safety; structured reasoning; agentic and multi-agent systems; optimisation-based inference; trustworthy AI; reasoning under uncertainty; interpretability and auditability; large language models; multimodal reasoning
Login
Register
Submit a Paper
Propose a Special lssue