Abstract

A system and method for the governed custody of data and machine-learning models, organized by sector to support Specialized SI (also known as Specialized AI), in which every deposit carries a content-derived identifier and machine-readable terms of use; every access by an AI model or software agent is mediated by a policy-enforcing regulator and executed at the location of the data; every use of a deposit in training is recorded against the resulting model version; a depositor may freeze or withdraw a deposit at any time, triggering an attested remediation obligation for every affected model; and verified restore points of models and data are held in a custody domain independent of the systems they restore, so that a system halted by an out-of-band reset can be restarted from a trusted state. Custody nodes are federated across operators and regions, so that data remains in place while models travel to it.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS