Introduction
Virginia’s legislature has advanced the High-Risk Artificial Intelligence Developer and Deployer Act (HB 2094) [6] [8], positioning the state as a leader in implementing comprehensive anti-discrimination measures against algorithmic bias. This legislation follows Colorado’s precedent and aims to regulate the use of high-risk AI systems in critical sectors.
Description
Virginia’s legislature has passed the High-Risk Artificial Intelligence Developer and Deployer Act (HB 2094) [6] [8], making it the second state to implement comprehensive anti-discrimination measures concerning algorithmic discrimination, following Colorado’s example from the previous year [5]. Introduced by Delegate Michelle Lopes Maldonado [2], the bill passed by a narrow margin in both the House of Delegates and the Senate, with the Senate voting 21-19 and the House approving it 51-47. It is currently awaiting the signature of Republican Governor Glenn Youngkin and, if enacted [4] [5] [7], will take effect on July 1, 2026 [4] [6]. There is potential for amendments before the bill’s implementation [4], as industry representatives have noted the importance of compliance and possible legislative adjustments [4].
The legislation specifically targets machine-learning-based AI systems that significantly influence consequential decisions in sectors such as education enrollment, employment access [6] [7], financial services [8], and healthcare [6] [7] [8]. It defines “high-risk artificial intelligence” as systems that autonomously make or are substantial factors in consequential decisions across various sectors [6]. A key aspect of the legislation is the establishment of a duty of reasonable care for businesses using automated decision-making systems [6], focusing on actively mitigating potential biases and preventing algorithmic discrimination based on protected characteristics. Notably, while the definition of “consumer” excludes individuals acting in a commercial or employment context [6], job applicants are still considered consumers when evaluated by AI-driven hiring tools [6].
HB 2094 sets a higher threshold for coverage compared to Colorado’s legislation [6], requiring that AI be the “principal basis” for decisions to trigger anti-discrimination provisions [6]. Developers of high-risk AI systems are mandated to provide comprehensive information about the system’s intended use [3], purpose [3] [6], known limitations [3] [6], potential risks of algorithmic discrimination [3], performance evaluation results [3], and risk mitigation strategies [3]. The law also requires that developers disclose information about intended uses and limitations [6], updating these disclosures following significant modifications [6]. Additionally, the legislation addresses concerns related to “deep fakes” by requiring detectable markings for AI-generated content and imposes transparency obligations for synthetic content generated by high-risk generative AI applications.
An economic analysis by Chamber of Progress estimates that compliance with this bill could impose costs of $290 million on Virginia’s AI innovators [2], which is significant given the state’s AI market is valued at $1.71 billion. This analysis suggests that AI developers typically allocate 17.22 percent of total model development costs for compliance [2], indicating that each model update could incur additional costs.
Both developers and deployers of high-risk AI systems must implement risk management policies and conduct impact assessments to identify potential discriminatory risks before deployment [6]. Deployers are required to inform consumers when AI systems will make consequential decisions [6], providing details about the system’s purpose and the nature of the decision [6]. In cases of adverse outcomes [6], such as job application rejections [6], deployers must disclose the reasons for the decision and allow individuals to correct inaccuracies or appeal [6]. The legislation includes a discretionary right to cure and stipulates that the state attorney general must issue a civil investigative demand (CID) prior to any enforcement actions [3].
Concerns have been raised about the potential impact of such regulations on Virginia’s status as an AI hub [2], with warnings that excessive regulation could drive innovation away [2]. The US Chamber of Commerce has expressed opposition to HB 2094, highlighting its potential impact on smaller businesses and its alignment with existing Virginia laws [1]. They advocate for a regulatory framework that builds on current laws before introducing new regulations [1], emphasizing that many AI activities are already covered by existing legislation [1]. The Chamber points out that the bill lacks provisions for small business relief [1], unlike the Virginia Consumer Data Protection Act [1], and notes that AI tools have been beneficial for small businesses [1], with a significant percentage already utilizing AI-enabled technologies [1]. Concerns are also raised about the increased litigation and compliance costs that small businesses may face due to a fragmented regulatory landscape [1], as many small companies are reportedly hesitant to adopt AI due to fears of legal complications [1]. Unlike Colorado’s law [8], the Virginia Act has a narrower scope and does not aim to broadly regulate AI usage [8], although other states, such as Connecticut [4], are pursuing similar regulatory measures [4].
Conclusion
The enactment of HB 2094 could significantly impact Virginia’s AI landscape, potentially increasing compliance costs and influencing the state’s attractiveness as an AI innovation hub. While the legislation aims to mitigate algorithmic discrimination, it also raises concerns about regulatory burdens on small businesses and the potential for stifling innovation. As other states consider similar measures, the balance between regulation and innovation remains a critical consideration for policymakers.
References
[1] https://www.uschamber.com/technology/u-s-chamber-letter-urging-governor-youngkin-to-veto-virginia-house-bill-2094
[2] https://progresschamber.org/new-analysis-ai-discrimination-bill-may-cost-virginia-developers-and-deployers-millions/
[3] https://www.transparencycoalition.ai/news/victory-in-virginia-the-old-dominion-approves-two-important-ai-bills
[4] https://privacy-daily.com/article/2025/02/20/va-legislature-passes-ai-bill-cdt-says-its-ineffective-2502200003?BC=bc_67b87e1352fc8
[5] https://www.multistate.ai/updates/vol-50
[6] https://natlawreview.com/article/charting-future-ai-regulation-virginia-poised-become-second-state-enact-law
[7] https://news.bgov.com/bloomberg-government-news/ai-anti-discrimination-legislation-passes-virginia-statehouse
[8] https://www.jdsupra.com/legalnews/virginia-legislature-passes-high-risk-6283689/
