According to the Legislative Budget Board (LBB), the fiscal implications of HB 2818 project a significant cost to the state over the next biennium and beyond. The bill is expected to have a negative impact of approximately $8.1 million on General Revenue Funds through August 31, 2027. This cost reflects the initial setup and ongoing operations of the newly mandated Artificial Intelligence Division within the Texas Department of Information Resources (DIR).

The largest cost drivers include the hiring of nine full-time equivalent employees (FTEs) and necessary technology investments. DIR plans to employ a mix of system analysts, data analysts, contract specialists, and project managers to build and manage the division’s AI infrastructure. Personnel costs are estimated at $3.64 million over the biennium, with approximately $1.81 million annually in the years following. These staff will be critical for tasks such as code development, project management, procurement coordination, and assessing AI risks like bias and data security.

On the technology side, the agency anticipates spending $1.5 million in fiscal year 2026 and $3 million in each following fiscal years. This includes costs for professional services such as AI assessments, data preparation, and prompt engineering, as well as cloud infrastructure to support the generative AI systems being implemented. These investments are necessary to support the state’s efforts to modernize outdated systems using advanced AI technologies.

Importantly, the fiscal note also clarifies that no significant impact is anticipated for local governments, as the scope of the bill is confined to state agencies and those entities using DIR services. Although the bill does not itself appropriate funds, it sets the legal framework for future appropriations to support this AI initiative.

Texas Policy Research recommends that lawmakers vote NO on HB 2812 unless amended as described below. This position reflects well-founded concerns about the bill’s fiscal impact and its expansion of state government without sufficient guardrails to ensure long-term efficiency and accountability. While the intent behind the bill—to modernize outdated government systems using generative artificial intelligence (AI)—is commendable, the mechanism it proposes raises red flags for those who prioritize limited government and fiscal discipline.

Specifically, the bill would create a new Artificial Intelligence Division within the Department of Information Resources (DIR), authorize the hiring of nine new state employees, and initiate a recurring cost of nearly $5 million annually after an initial $8.1 million through 2027​. While the division is tasked with identifying efficiencies, no clear statutory limits, performance metrics, or sunset provisions are included to ensure that the division actually delivers on its cost-saving promises. This lack of built-in accountability makes it difficult to justify the expansion of government infrastructure and ongoing taxpayer obligations at this scale.

Moreover, the bill does not place any new regulatory burdens on individuals or businesses, nor does it extend government power into private markets. However, by building a permanent in-house capability rather than leveraging competitive private-sector solutions, it risks duplicating functions that could be handled more efficiently outside of government.

Suggested Amendments:

• Add a sunset provision requiring legislative reauthorization after a fixed term.
• Include performance benchmarks to ensure the division meets specific cost-saving or efficiency goals.
• Impose budgetary caps and staffing limits to contain long-term costs.
• Mandate public transparency in cost analysis reporting.
• Explore a public-private partnership model as an alternative to expanding state infrastructure.

In conclusion, the bill is conceptually forward-thinking but structurally underdeveloped. With amendments—such as performance-based triggers, sunset review, capped budgets, and private-sector collaboration requirements—the legislation could better align with conservative principles of government restraint.

• Individual Liberty: The bill does not directly impact Texans’ civil liberties, as it focuses on internal state agency operations. However, because it involves the use of generative AI, which may process large amounts of data or automate decision-making, future applications could potentially affect public interactions with government services. Without strong transparency or oversight requirements, there's a latent risk that AI could erode trust in fair and accountable public systems if misused. That said, this risk is theoretical under the current scope of the bill.
• Personal Responsibility: The bill supports this principle in theory by encouraging state agencies to modernize outdated systems and be better stewards of public resources. By mandating cost analysis reports, the bill promotes a culture of evaluating outcomes and justifying investments. However, it falls short of requiring that these analyses lead to corrective actions, which would strengthen the accountability of agencies for results, an essential aspect of personal responsibility in government.
• Free Enterprise: The bill is a mixed bag for free enterprise. On the positive side, it permits DIR to contract with private AI vendors, potentially creating business opportunities in Texas's growing technology sector. However, by also building a permanent state-run AI division, the government may be positioning itself as a competitor to private firms over time, especially in areas like software development or system integration. This could unintentionally disincentivize private-sector innovation if not carefully constrained or guided by a market-friendly procurement policy.
• Private Property Rights: There are no direct intrusions on property rights in the traditional sense (land or possessions), but in the modern digital landscape, data can be viewed as a form of property. Since the bill involves the use of generative AI, which often requires large data sets, there’s a need for safeguards to prevent inappropriate data collection or use, even if those issues are not addressed in the current version. To fully respect evolving concepts of digital property, stronger privacy and data usage controls could be added.
• Limited Government: This is where the bill most clearly conflicts with liberty principles. It grows the size of the state government by creating a new division, adding staff, and establishing a new budget line item with no built-in end date or performance-based review. While the goal is to make the government more efficient, this is done by first expanding it. That may be a contradiction in terms unless the bill is amended to include limits such as a sunset clause, performance benchmarks, or clear return-on-investment requirements.

• SB 1700 (Same As)