As the 2024 elections approach and the window for Congress to consider bipartisan comprehensive artificial intelligence (AI) legislation shrinks, California officials are attempting to guard against a generative AI free-for-all—at least with respect to state government use of the rapidly advancing technology—by becoming the largest state to issue rules for state procurement of AI technologies. Without nationwide federal rules, standards set by state government procurement rules may ultimately add another layer of complexity to the patchwork of AI-related rules and standards emerging in the states.
On March 21, the California Government Operations Agency (GovOps) published interim guidelines for government procurement of generative AI technologies. The new guidance directs state officials responsible for awarding and managing public contracts to identify risks of generative AI, monitor the technology’s use, and train staff on acceptable use, including for procurements that only involve “incidental” AI elements. For “intentional” generative AI procurements, where an agency is specifically seeking to purchase a generative AI product or service, the guidelines impose a higher standard: in addition to the requirements that apply to “incidental” purchases, agencies seeking generative AI technologies are responsible for articulating the need for using generative AI prior to procurement, testing the technology prior to implementation, and establishing a dedicated team to monitor the AI on an ongoing basis.
The interim guidance follows the executive order Governor Gavin Newsom (D) signed in September to require state agencies to study, regulate, and otherwise prepare for implementation of generative AI for both government and broader public use in California. The state expects to finalize the procurement policy in 2025.
The federal government is in the initial stages of promulgating its own AI guidance under President Biden’s executive order. As our colleagues have explained, the Office of Management and Budget (OMB) issued guidance to federal agencies on March 28 outlining principles and best practices for responsible AI use and concurrently requested public comment on its approach. Although Congress has not yet passed comprehensive AI legislation this term, several targeted bills have moved through the Senate, and additional bills are building momentum in the House, driven in large part by bipartisan AI working groups in both chambers.