Relating to the artificial intelligence division within the Department of Information Resources.
ModeratePlan for compliance
Low Cost
Effective:2025-06-20
Enforcing Agencies
Department of Information Resources (DIR)
01
Compliance Analysis
Key implementation requirements and action items for compliance with this legislation
Implementation Timeline
Effective Date:June 20, 2025 (Immediate effect due to supermajority vote).
Compliance Deadline:Immediate for new solicitations. Any proposal submitted to DIR for legacy system modernization after June 20, 2025, is subject to the "majority work" AI requirement.
Agency Rulemaking: DIR is mandated to adopt rules "as soon as practicable." Expect a regulatory gray zone between now and Q4 2025 where DIR operates based on internal guidance before formal rules are published in the *Texas Register*.
Immediate Action Plan
1.Audit Active Pipelines: Review all pending bids for Texas IT modernization. If you proposed a manual labor model, you must pivot immediately to an AI-assisted model to remain compliant.
2.Define "Baseline" Methodologies: Create a standard internal formula for calculating "Traditional Costs." You will need this to defend your "Savings" numbers in the mandatory post-project reports.
3.Update IP Policies: Verify that your GenAI tools do not infringe on third-party IP. The state will transfer this risk entirely to you via contract indemnification.
4.Monitor the Texas Register: Assign a compliance officer to watch for DIR's proposed rules regarding the definition of "majority work" and security standards for AI code.
Operational Changes Required
Contracts
"Majority Work" Clauses: New State of Texas contracts for modernization will mandate that GenAI completes a "majority" of the work.
*Critical Note:* While the statute says "majority" (>50%), the Legislative Budget Board's Fiscal Note explicitly assumes 90% of the work will be AI-generated. Expect DIR contract negotiators to push for the 90% threshold, not 51%.
IP Indemnification: You must update Master Services Agreements (MSAs) to address AI-generated code. The state will likely require vendors to indemnify the state against copyright claims arising from the training data used by the vendor's AI tools.
Hiring/Training
New Role Requirements: Your proposal teams must now include "AI Suitability Assessments." You cannot bid on these projects without a technical lead capable of certifying which portions of the legacy code are compatible with GenAI ingestion.
Staffing Shift: The Fiscal Note anticipates DIR hiring internal Systems Analysts to evaluate AI code security and bias. Your delivery teams must be trained to interface with these specific state oversight roles, particularly regarding "Human-in-the-Loop" (HITL) security verification.
Reporting & Record-Keeping
Mandatory Cost Analysis Report: Sec. 2054.703 creates a statutory requirement to file a report upon project completion.
Required Metrics: You must track and report three specific data points to get paid/close out the contract:
1.Time Saved: AI delivery vs. estimated manual delivery.
2.Money Saved: Hard currency comparison.
3.Resource Efficiency: FTE hours saved.
Baseline Data: You must establish an agreed-upon "manual baseline" cost *before* the project starts; otherwise, you cannot legally calculate the "savings" required for the final report.
Fees & Costs
No New Statutory Fees: There are no new filing fees.
Tooling Costs: Vendors bear the cost of the GenAI licenses and compute power unless the contract specifies the use of a state-provided environment.
Strategic Ambiguities & Considerations
Definition of "Work": The statute requires AI to complete the "majority of the work" but fails to define the metric. Is this calculated by Lines of Code (LOC), Function Points, or Billable Hours?
*Risk:* If DIR defines this by LOC, it incentivizes bloated AI code. If by hours, it incentivizes efficiency. Watch the rulemaking closely for this definition.
Vendor vs. State AI Stack: Sec. 2054.702(b) allows DIR to use its own AI or contract a vendor's AI. It is unclear if vendors will be forced to use a specific, state-approved AI platform (e.g., a specific Azure/OpenAI instance) or if vendors can bring their own proprietary tools.
Liability for "Hallucinations": The law is silent on liability for errors introduced by the mandated AI. Until rules are set, vendors should assume strict liability for code security flaws, even if the AI generated them.
Need Help Understanding Implementation?
Our government affairs experts can walk you through this bill's specific impact on your operations.
Information presented is for general knowledge only and is provided without warranty, express or implied. Consult qualified government affairs professionals and legal counsel before making compliance decisions.
The author has informed the committee that, traditionally, the Department of Information Resources (DIR) contracts for legacy modernization projects or other projects requested by state agencies through a procurement process, and the cost to complete these projects sometimes exceeds the amount of money an agency has budgeted due in part to the number of people needed to complete a project. C.S.H.B. 2818 seeks to assist state agencies in the modernization of legacy systems and the completion of other projects in a more efficient manner by establishing the Artificial Intelligence Division within DIR to oversee and support the use of generative artificial intelligence technologies in state projects.
CRIMINAL JUSTICE IMPACT
It is the committee's opinion that this bill does not expressly create a criminal offense, increase the punishment for an existing criminal offense or category of offenses, or change the eligibility of a person for community supervision, parole, or mandatory supervision.
RULEMAKING AUTHORITY
It is the committee's opinion that rulemaking authority is expressly granted to the Department of Information Resources in SECTION 1 of this bill.
ANALYSIS
C.S.H.B. 2818 amends the Government Code to require the Department of Information Resources (DIR) to establish an artificial intelligence division within DIR and to require DIR's executive director to direct the administration and oversight of the division, including the division's organization within DIR. The bill authorizes DIR to hire the personnel necessary to administer the division's duties, including hiring personnel through public institutions of higher education.
C.S.H.B. 2818 requires the division to assist state agencies and other entities that use DIR's services in the implementation of generative artificial intelligence (AI) technology for the following:
·projects to modernize or replace legacy systems as defined under the Information Resources Management Act; and
·other projects appropriate for the use of generative AI technology as determined by DIR.
C.S.H.B. 2818 authorizes the division to do the following in assisting agencies and entities with the applicable projects:
·use generative AI technology developed or procured by DIR to complete the project; or
·contract with a vendor to use generative AI technology to complete the project, in which case the vendor's generative AI technology must complete a majority of the work required to complete the contract.
C.S.H.B. 2818 requires the division to prepare a cost analysis report for each project completed under the bill's provisions that includes the following:
·a summary of the time, money, and resources saved by using a generative AI technology compared to the time, money, and resources that would have been required using traditional systems to perform equivalent tasks; and
·a concise overview focused on information demonstrating the cost and time efficiencies achieved by using generative AI technology for the project.
C.S.H.B. 2818 authorizes DIR to adopt the rules necessary to establish and administer the division and requires DIR to adopt the rules necessary to implement the bill's provisions as soon as practicable after the bill's effective date.
EFFECTIVE DATE
September 1, 2025.
COMPARISON OF INTRODUCED AND SUBSTITUTE
While C.S.H.B. 2818 may differ from the introduced in minor or nonsubstantive ways, the following summarizes the substantial differences between the introduced and committee substitute versions of the bill.
The substitute lowers the threshold of the amount of work that the generative AI technology of a vendor with which the division contracts to assist agencies and entities with applicable projects is required to complete from at least 90 percent of the work required to complete the contract, as in the introduced, to a majority of such work.
The substitute revises the provision of the introduced requiring a project cost analysis report to contain a concise overview in the following ways:
·omits the specification that the overview information includes the amount of time and money saved in the procurement process, which was present in the introduced; and
·includes information demonstrating the cost and time efficiencies achieved by using generative AI technology for the project, which did not appear in the introduced.
Honorable Giovanni Capriglione, Chair, House Committee on Delivery of Government Efficiency
FROM:
Jerry McGinty, Director, Legislative Budget Board
IN RE:
HB2818 by Capriglione (Relating to the artificial intelligence division within the Department of Information Resources.), As Introduced
Estimated Two-year Net Impact to General Revenue Related Funds for HB2818, As Introduced: a negative impact of ($8,140,645) through the biennium ending August 31, 2027.
The bill would make no appropriation but could provide the legal basis for an appropriation of funds to implement the provisions of the bill.
General Revenue-Related Funds, Five- Year Impact:
Fiscal Year
Probable Net Positive/(Negative) Impact to General Revenue Related Funds
2026
($3,333,999)
2027
($4,806,646)
2028
($4,806,646)
2029
($4,806,646)
2030
($4,806,646)
All Funds, Five-Year Impact:
Fiscal Year
Probable Savings/(Cost) from General Revenue Fund 1
Change in Number of State Employees from FY 2025
2026
($3,333,999)
9.0
2027
($4,806,646)
9.0
2028
($4,806,646)
9.0
2029
($4,806,646)
9.0
2030
($4,806,646)
9.0
Fiscal Analysis
The bill would require the Department of Information Resources (DIR) to establish an artificial intelligence (AI) division, authorizing DIR to hire, including through institutions of higher education, personnel necessary to administer the duties of the division.
The bill would require the division to assist state agencies and other entities that use DIR's services to implement generative AI technology for modernization projects and other projects appropriate for the use of generative AI technology as determined by DIR. It would authorize DIR to use generative AI to complete the projects. It would also authorize DIR to contract with a vendor to use generative AI technology, requiring that 90 percent of the work be completed by the vendor's generative AI technology.
The bill would require the division to prepare a cost analysis report for each project completed that includes a summary of time, money, and resources saved and cost efficiencies achieved by using generative AI technology.
DIR would adopt rules as necessary to implement the legislation.
Methodology
It is assumed that DIR would require 9.0 additional FTEs as follows to implement the legislation: 5.0 Systems Analyst VIIs to train, support coding efforts, help craft statements of work and evaluations, and understand risk components of code security, bias, and privacy; 1.0 Data Analyst V to to act as a data research specialist knowledgeable in AI tools and data management; 1.0 Contract Specialist V to support the procurement and contract coordination between DIR's Shared Technology Services and Cooperative Contracts Program, and other state agencies; and 2.0 Program Management Specialist IIIs to act as project management/technical writers knowledgeable in agile application development practices. Total personnel costs would be $3,640,645 for the biennium, and $1,806,646 in each subsequent fiscal year.
Technology
DIR estimates the total technology cost would be $1,500,000 in fiscal year 2026 and $3,000,000 in each subsequent fiscal year. This includes $1,000,000 in fiscal year 2026 and $2,000,000 in each subsequent fiscal year in professional service costs for AI assessments, data analysis and cleansing, model testing, and prompt engineering. It also includes $500,000 in fiscal year 2026 and $1,000,000 in each subsequent fiscal year for cloud services to establish generative AI environments to support various technologies.
Local Government Impact
No significant fiscal implication to units of local government is anticipated.
Source Agencies:
313 Department of Information Resources
LBB Staff:
JMc, RStu, LCO, CSmi, NV
Related Legislation
Explore more bills from this author and on related topics
HB2818 fundamentally alters the procurement model for state IT modernization projects, shifting the preference from human-labor-intensive coding to Generative AI (GenAI) automation. Effective immediately, vendors bidding on legacy system replacement contracts with the Department of Information Resources (DIR) must demonstrate that GenAI will perform the majority of the work, creating a new barrier to entry for firms relying on traditional manual development models. Implementation Timeline Effective Date: June 20, 2025 (Immediate effect due to supermajority vote).
Q
Who authored HB2818?
HB2818 was authored by Texas Representative Giovanni Capriglione during the Regular Session.
Q
When was HB2818 signed into law?
HB2818 was signed into law by Governor Greg Abbott on June 20, 2025.
Q
Which agencies enforce HB2818?
HB2818 is enforced by Department of Information Resources (DIR).
Q
How urgent is compliance with HB2818?
The compliance urgency for HB2818 is rated as "moderate". Businesses and organizations should review the requirements and timeline to ensure timely compliance.
Q
What is the cost impact of HB2818?
The cost impact of HB2818 is estimated as "low". This may vary based on industry and implementation requirements.
Q
What topics does HB2818 address?
HB2818 addresses topics including artificial intelligence, purchasing, purchasing--state, state agencies, boards & commissions and information resources, department of.
Legislative data provided by LegiScanLast updated: November 25, 2025
Need Strategic Guidance on This Bill?
Need help with Government Relations, Lobbying, or compliance? JD Key Consulting has the expertise you're looking for.
