OIT says it maintains AI inventory as lawmakers demand clearer mapping to SB24-205 consequential decisions

The Office of Information Technology (OIT) told the Joint Technology Committee that it maintains an inventory of AI systems and is seeking resources to support statewide compliance with Senate Bill 24-205, but lawmakers pressed the agency for a narrowed, consumer-facing inventory and for a mapping of systems that make consequential decisions.

David Edinger, executive director and chief information officer for the state, introduced new OIT leadership and described an organizational realignment meant to accelerate delivery. Ross Pascal, OIT’s budget director, summarized R1 (Statewide AI Compliance) as a request for dedicated FTEs that would sit inside agencies to analyze, assess and monitor AI systems. "These are for dedicated FTE resources that would reside within each agency to perform the necessary functions to implement the requirements as currently stated in Senate Bill 24-205," Pascal said, identifying an initial request for about 33.8 FTE in year one and 36.9 FTE in a subsequent year.

Lawmakers repeatedly asked for a clearer list of systems that would be subject to SB24-205 and for a column that identifies whether a system is customer-facing and whether it performs a consequential decision. "We do have an inventory of all of the AI systems," Edinger said, "and we are tracking that. And we can share that." He added that OIT’s current risk-assessment process is based on the NIST framework and that the agency used a methodology separate from what might be required under SB24-205 because the bill’s implementing guidance and working-group recommendations are still pending.

Committee members said the fiscal note and staffing request need to be grounded in an inventory that distinguishes internal administrative AI from consumer-facing systems that could trigger SB24-205 compliance. Representative Pascall asked that departments identify the "consequential decision" the software makes so the committee can determine which systems fall under the statute. Senator Rodriguez and others said they want OIT and JBC staff to coordinate on questions to avoid duplicative requests and to return with a narrowed dataset.

OIT estimates its inventory contains roughly 200 distinct generative-AI technologies in use across state agencies, many of which are internal tools that are not consumer-facing. Edinger said the AI working group is close to issuing recommendations and that OIT will reassess fiscal needs based on those outcomes. Committee members directed OIT to provide a consumer-facing inventory and to work with JBC staff; the committee will reconvene to continue its review of R1 and related requests after those steps.

The exchange underscores the legislature’s demand that funding for AI compliance be tightly tied to concrete evidence of which systems will require oversight and whether those systems make consequential decisions affecting Coloradans.