Google introduces Simula for synthetic data generation

Google researchers introduced Simula, a framework that reframes synthetic data generation as dataset-level mechanism design. Simula uses reasoning to architect datasets from first principles, enabling fine-grained control over coverage, complexity, and quality for privacy-sensitive or data-scarce domains. The paper was published in Transactions on Machine Learning Research.