Vladislav Yurukov, Faculty of Informatics, South-West University "Neofit Rilski", Bulgaria (in person) video
Effective guardrails for artificial intelligence (AI) require a multi-dimensional approach that integrates governance, transparency, safety, accountability, and human oversight across the AI life cycle. Drawing on contemporary perspectives from AI governance, clinical safety, health informatics, and ethics literature, best practices emphasize: (1) explicit governance frameworks and risk management tailored to use-case risk, (2) rigor in transparency, explainability, and documentation, (3) robust safety guardrails including monitoring for drift and mis- and disinformation, (4) fairness, privacy, and inclusivity in data and model behavior, (5) human-in-the-loop supervision and auditability, and (6) ongoing regulatory alignment and multidisciplinary engagement. Below, each facet is articulated, with cross-referenced evidence and practical recommendations for implementation. Where opinions diverge among sources, nuances are noted and areas needing further consensus.