Everett Thornton Smith

RAND expert insights piece proposing best-practice approaches and a standards taskforce for GPAI model evaluation in the EU; Everett Smith is listed among the authors/contributors.

Working paper / workshop submission proposing desiderata and a proposed EU taskforce to set standards for GPAI (general-purpose AI) evaluations (internal validity, external validity, reproducibility, portability).

Multi-author report/catalog enumerating open technical research problems relevant to AI governance, covering topics from data and compute to evaluation and deployment; Everett Thornton Smith is listed among contributors/authors.

Paper describing requirements and an ecosystem for effective external scrutiny (red‑teaming, auditing, external researcher access) of frontier large language models; introduces the ASPIRE framework (Access, Searching attitude, Proportionality, Independence, Resources, Expertise).