LLMs and toxicology: OpenTox virtual conference

At the upcoming OpenTox Virtual Conference 2025, experts Arianna Bassan and Glenn Myatt will highlight how Large Language Models (LLMs) can support toxicological risk assessment while ensuring accuracy and regulatory trust.

Glenn Myatt, in his talk “Frameworks for Using Large Language Models in Toxicological Risk Assessment,” presents a structured framework for applying LLMs to evaluate chemical exposure and biological targets. Developed with input from an industry consortium and tested through case studies, the framework tackles known LLM challenges (such as inconsistent outputs and lack of transparency) and discusses how AI can assist in every stage of the assessment process, from data capture to evidence interpretation.

Arianna Bassan’s presentation, “Regulatory Acceptance and Use of Large Language Models in Toxicological Risk Assessment,” emphasizes the importance of accuracy, transparency, and accountability when integrating LLMs into safety evaluation workflows. She advocates a human-in-the-loop approach, where experts supervise AI outputs for data extraction, evidence summarization, and report drafting. Arianna argues that this supervision preserves traceability and credibility, ensuring that toxicologists remain the final decision-makers and that LLM-assisted assessments are both defensible and reproducible.

Together, Glenn and Arianna’s work underscores a shared vision: AI tools can enhance toxicological science when used under expert review and within transparent, well-defined frameworks, ensuring both efficiency of the process and reliability of the outcome.