On March 24, 2025, Virginia Governor Glenn Youngkin vetoed HB 2094 on High-Risk Artificial Intelligence Systems, a bill that passed through the state legislature by slim margins and which would have established a regulatory regime largely mirroring last year’s Colorado AI Act.
As with the Colorado Act, HB 2094 would have imposed a number of requirements on organizations that develop or deploy high-risk AI systems that make certain consequential decisions. Such requirements would have included the completion of impact assessments, maintaining technical documentation, adopting risk management frameworks, and providing individuals with a means to request review of adverse decisions undertaken by AI.
In his veto message, Governor Youngkin cited concern that HB 2094 “would undermine [ ] progress, and risks turning back the clock on Virginia’s economic growth, stifling the AI industry as it is taking off.” The message also points to existing laws on discriminatory practices, privacy, data use, and libel that can already be harnessed to protect the public from potential AI harms.
The governor’s veto underscores the challenges that U.S. legislators and policymakers have faced in striking a balance between encouraging innovation while promoting the responsible, safe development of AI. For instance, Texas lawmakers recently shelved their proposed Texas Responsible AI Governance Act for a new bill that focuses more narrowly on discrimination against protected classes and government use of AI. Last year, California Governor Gavin Newsom vetoed the enactment of the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act following a tepid response from the industry. The Colorado AI Act which remains the only U.S. law on the books to comprehensively govern AI, is also slated for significant revisions.
With the Trump administration retreating from more proactive regulatory approaches adopted by the predecessor administration, some state lawmakers are expected to step up. But HB 2094 veto highlights the fraught path to comprehensive AI regulation at the state level and significant foundational work still needed to achieve consensus around how to regulate AI.
