AI systems are sensitive to data: training data, model states, inference results. Walrus is also created to support large data blobs, which happen to be the natural fit with AI processes that need scale, integrity and reproducibility.
Walrus supports onchain storage and verification of AI models and datasets as well as referencing. This facilitates auditable AI pipelines in which the results may be tracked to precise sources of data. Walrus does not run AI it provides the data base that enables decentralized and trusted AI.
