Texas House of Representatives · 90th Legislature, Regular Session
Official title: Relating to human oversight and override for certain high-risk artificial intelligence systems; providing a civil penalty.
Members who authored, joint-authored, or co-sponsored this bill.
SourceLegiScan (CC BY 4.0)90R
The bill adds Subchapter D to Chapter 552, Business & Commerce Code, which covers developers and deployers of a "high-risk artificial intelligence system," defined as an AI system that can independently start or carry out an action posing a reasonably foreseeable risk of substantial harm to people's health or safety, property, critical infrastructure, or the security of an information system or confidential information. A developer who knows or reasonably should know its system will be used this way shall provide human oversight and override mechanisms and give the deployer the documentation reasonably necessary to install, configure, maintain, and test them. The developer shall also test those mechanisms before the system's initial deployment and after a material change that could affect them. A deployer shall maintain mechanisms that let an authorized human operator (a person the deployer designates who has the needed training, authority, and technical access) monitor the system and approve, reject, modify, override, or terminate its decisions and actions. The deployer shall test those mechanisms before initial deployment, at least annually, after a material change, and after a material failure before resuming affected operations. A developer or deployer may not suspend, terminate, or take other adverse personnel action against an employee, former employee, or contractor who in good faith reports a suspected violation to the attorney general or the Department of Information Resources, or who assists an investigation. The attorney general enforces the subchapter under Subchapter C, which is not provided here, so the civil penalty amount is not stated in this text, and the bill takes effect September 1, 2027.
Read the full text below for the definitive meaning.
Each version links to the authoritative document on capitol.texas.gov.
By: Bhojani H.B. No. 510 A BILL TO BE ENTITLED AN ACT relating to human oversight and override for certain high-risk artificial intelligence systems; providing a civil penalty. BE IT ENACTED BY THE LEGISLATURE OF THE STATE OF TEXAS: SECTION 1. Chapter 552, Business & Commerce Code, is amended by adding Subchapter D to read as follows: SUBCHAPTER D. HUMAN OVERSIGHT AND OVERRIDE OF HIGH-RISK ARTIFICIAL INTELLIGENCE SYSTEMS
Sec. 552.151. DEFINITIONS. In this subchapter: (1) "Authorized human operator" means an individual designated by a deployer who possesses the training, authority, and technical access necessary to: (A) monitor a high-risk artificial intelligence system; and (B) approve, reject, modify, override, and terminate, as applicable, the decisions, proposed actions, or ongoing actions of a high-risk artificial intelligence system. (2) "Department" means the Department of Information Resources. (3) "High-risk artificial intelligence system" means an artificial intelligence system, as defined by Section 551.001, that is capable of independently initiating or executing an action that presents a reasonably foreseeable risk of substantial harm to: (A) the health or safety of one or more individuals; (B) property; (C) critical infrastructure, as defined by Section 421.001, Government Code; or (D) the security of an information system or confidential information. Sec. 552.152. APPLICABILITY. (a) This subchapter applies to a developer, as defined by Section 552.001, who knows or reasonably should know that an artificial intelligence system developed by the developer will be used as a high-risk artificial intelligence system. (b) This subchapter applies to a deployer, as defined by Section 552.001, who deploys or operates a high-risk artificial intelligence system. Sec. 552.153. DEVELOPER RESPONSIBILITIES. (a) A developer subject to this subchapter shall: (1) provide human oversight and override mechanisms capable of enabling a deployer to comply with Section 552.154; (2) test the mechanisms before the initial deployment of the system; (3) test the mechanisms following a material change to the system that could affect the effectiveness of those mechanisms; and (4) provide the deployer with documentation reasonably necessary to install, configure, maintain, and test the mechanisms and comply with this subchapter. (b) A developer who has complied with the requirements of this subchapter is not liable for a violation of this subchapter solely because a deployer fails to install, enable, maintain, or properly use a mechanism provided by the developer. Sec. 552.154. DEPLOYER RESPONSIBILITIES. (a) A deployer of a high-risk artificial intelligence system shall implement and maintain human oversight and override mechanisms that enable an authorized human operator to: (1) monitor the system's operations; and (2) approve, reject, modify, override, and terminate, as applicable, the decisions, proposed actions, or ongoing actions of the system. (b) A deployer shall implement and maintain mechanisms reasonably designed to prevent the system or an unauthorized person from circumventing, disabling, or interfering with the mechanisms required by Subsection (a). (c) A deployer shall implement and maintain additional mechanisms beyond the mechanisms required by Subsection (a) to an extent and frequency that is proportionate to the nature and risk of the decision or action of the system, considering: (1) the severity of potential harm; (2) the system's degree of autonomy; and (3) the ability of a human operator to intervene before harm occurs. (d) A deployer shall test the mechanisms required by this section: (1) before initial deployment; (2) at least annually after initial deployment; (3) following a material change to the system that could affect the effectiveness of those mechanisms; and (4) following a material failure of those mechanisms, before resuming affected operations. Sec. 552.155. PROHIBITION ON RETALIATION. A developer or deployer may not suspend or terminate the employment or contract of, or take other adverse personnel action against, an employee, former employee, or contractor who in good faith: (1) reports conduct the individual believes violates this subchapter to the attorney general or the department; or (2) provides information or otherwise assists in an investigation or proceeding concerning an alleged violation of this subchapter. Sec. 552.156. ENFORCEMENT. (a) A violation of this subchapter is subject to enforcement by the attorney general under Subchapter C. (b) Notwithstanding Section 552.104(b), the attorney general may bring an action seeking temporary injunctive relief before providing the notice required by Section 552.104(a) if continued operation of a high-risk artificial intelligence system in violation of this subchapter presents an immediate and substantial risk of serious harm. The attorney general shall provide notice under Section 552.104(a) as soon as practicable after bringing the action. (c) Notwithstanding Section 552.102, an employee, former employee, or contractor of a developer or deployer may submit a complaint alleging a violation of this subchapter through the online mechanism maintained under Section 552.102. (d) In determining whether a developer or deployer has complied with the technical requirements of this subchapter, the attorney general shall consult with the department and provide the department with relevant reports and technical documentation reasonably necessary for that evaluation. (e) Information obtained by the attorney general or the department in connection with an investigation or action under this subchapter, including the identity of an individual who reports an alleged violation of this subchapter, is confidential and not subject to disclosure under Chapter 552, Government Code, except that the information may be used in an action brought under this chapter. The provision of information by the attorney general to the department under Subsection (d) does not require notice or consent under Section 15.10(i). (f) This section does not authorize the department to bring an enforcement action under this chapter. SECTION 2. This Act takes effect September 1, 2027.
Each subject links to other 90R bills tagged the same way.
SourceLegiScan (CC BY 4.0)subject tags
SourceTexas Legislative Council (capitol.texas.gov)90R
No recorded roll-call votes on this bill.
SourceLegiScan (CC BY 4.0)90R
No fiscal notes or analyses recorded for this bill.
SourceLegiScan (CC BY 4.0)official fiscal notes & analyses (TLC)
Perch never collects comments itself — it points you to the Legislature's own hearing schedule and official public-comment channels (ADR-0007).
Official hearings & comment (capitol.texas.gov) →SourceTexas Legislative Council (capitol.texas.gov)official bill text (capitol.texas.gov)