WHAT THIS BILL REGULATES · 1 REQUIREMENT TYPE
How Is This Bill Enforced
Verbatim statutory text on the left; plain-language analysis and a per-section checklist on the right. Numbered markers cross-link to the matching checklist row.
(q) 1 Except as provided by Subsection (r), the agency shall prohibit the use of artificial intelligence, including an artificial intelligence-based algorithm, algorithmic device, automated scoring engine, or any scoring method that uses artificial intelligence in part, to score a constructed response on an assessment instrument administered under this section or an assessment system developed under Section 39.027(e). This subsection does not prohibit the use of a machine to score a multiple choice response.
(r) 2 A scoring method otherwise prohibited by Subsection (q) may be used to score a constructed response described by that subsection if the scoring method: (1) was trained on representative samples of constructed responses that include the responses of students who are: (A) educationally disadvantaged; (B) emergent bilingual students, as defined by Section 29.052; or (C) eligible to enroll in a school district's special education program under Subchapter A, Chapter 29; (2) demonstrates scoring validity and reliability at levels consistent with industry-accepted standards for automated machine or hybrid automated machine scoring methods, including standards established by the National Assessment of Educational Progress; and (3) has been evaluated by a qualified independent entity, unaffiliated with the agency, an advisory group of the agency, or any entity involved in the development of the scoring method and the entity certifies that the scoring method: (A) is valid and reliable in scoring constructed responses; (B) does not demonstrate any measurable bias against a student described by Subdivision (1); and (C) meets all applicable psychometric standards for the scoring of assessment instruments.
This section creates a default prohibition on the use of artificial intelligence to score constructed responses on state assessment instruments, while preserving machine scoring of multiple-choice items. The prohibition is broad, covering AI-based algorithms, automated scoring engines, and any scoring method that uses AI even in part.
The conditional exception in subsection (r) permits AI-based scoring only when three cumulative safeguards are met: the training data must be representative of disadvantaged student populations (educationally disadvantaged, emergent bilingual, and special-education-eligible students); the method must meet industry-accepted validity and reliability standards including NAEP benchmarks; and a qualified independent entity — unaffiliated with TEA, its advisory groups, or the scoring method developer — must certify validity, absence of measurable bias, and compliance with psychometric standards. The independent-evaluation requirement functions as a mandatory pre-deployment audit that cannot be satisfied by the agency or its contractors.
This Act applies beginning with the 2025-2026 school year.
Specifies that the Act's requirements apply beginning with the 2025–2026 school year, establishing the operational timeline for TEA to implement the prohibition and conditional exception for AI scoring of constructed responses.
This Act takes effect immediately if it receives a vote of two-thirds of all the members elected to each house, as provided by Section 39, Article III, Texas Constitution. If this Act does not receive the vote necessary for immediate effect, this Act takes effect September 1, 2025.
Provides that the Act takes effect immediately upon receiving a two-thirds supermajority vote in each chamber, as authorized by Article III, Section 39 of the Texas Constitution. Absent such a vote, the Act takes effect on September 1, 2025.