While improving human livelihoods, chemicals can be released into the environment, becoming emerging pollutants, and posing risks to human and ecological health. Given the vast number of chemicals, reliance solely on experimental methods for measuring their exposure and hazard parameters is impractical. The challenge has driven rapid advancements in computational toxicology. Traditional computational methods, such as molecular simulation and quantitative structure-activity relationship models, can be employed to predict parameters related to environmental exposure, hazards, and risks. Additionally, force fields based on machine learning can increase the speed of molecular simulations while maintaining modeling accuracy. Machine learning-based screening models enhance the identification of hazardous chemicals. Integrating advanced AI techniques into computational toxicology significantly improves chemical risk assessment and control of hazardous chemicals. Advancements in multimodal learning enabled the combination of multimodal data into computational models. Beyond assessing chemical risks, generative AI technologies provide innovative solutions for the molecular design of green alternative chemicals. Multi-constraint molecular generation models ensure that the designed chemicals maintain necessary functionalities and exhibit low hazards. The application of AI in computational toxicology is poised to minimize the adverse impacts of chemicals at the source.

Computational Toxicology,Chemicals,Emerging Pollutants,Risk Control