Texas AI hiring law: what TRAIGA actually requires of employers

Texas's AI law for employers, the Texas Responsible Artificial Intelligence Governance Act (TRAIGA, from House Bill 149), took effect January 1, 2026. It does not require private employers to disclose AI use to applicants, does not mandate bias audits or impact assessments, and gives no candidate a private right to sue. What it does is prohibit developing or deploying an AI system with the intent to unlawfully discriminate against a protected class, and it gives enforcement exclusively to the Texas Attorney General.

What TRAIGA is, and what it replaced

Governor Greg Abbott signed HB 149 on June 22, 2025; it took effect January 1, 2026, per Section 10 of the enrolled bill, available on capitol.texas.gov. It creates Chapter 551 (general provisions) and Chapter 552 (the AI protection chapter) of the Texas Business and Commerce Code.

HB 149 replaced an earlier, more sweeping proposal, HB 1709, that would have imposed mandatory algorithmic impact assessments, applicant-facing transparency disclosures and vendor accountability duties on employers using AI in hiring. None of that survived into the enacted law. The employment-specific text that made it into TRAIGA is a single discrimination provision, not a hiring-process statute.

The employment provision: intentional discrimination only

Section 552.056 states that a person "may not develop or deploy an artificial intelligence system with the intent to unlawfully discriminate against a protected class" and adds, in the same section, that "a disparate impact is not sufficient by itself to demonstrate an intent to discriminate." Two things follow for a hiring tool:

• Intent is the trigger. Unlike California's FEHA regulations or Title VII, which reach a facially neutral tool that produces a discriminatory result, TRAIGA's own provision targets a system built or used with discriminatory intent. A tool that produces an uneven outcome across protected groups, without more, does not by itself establish a TRAIGA violation.
• "Protected class" is not separately defined in Chapter 552; TRAIGA points to protections "under state or federal law," which pulls in Texas Labor Code Chapter 21 and federal anti-discrimination statutes rather than creating a new list.

Chapter 552 also carries general provisions against AI used to manipulate human behavior in a way that causes harm and against certain social-scoring uses by government; those are not employment-specific and are outside what a recruiter needs to track for hiring.

Enforcement: Attorney General only, with a cure period

Section 552.101 gives the attorney general "exclusive authority to enforce this chapter." There is no private right of action; an employee cannot bring a TRAIGA claim directly. Before the attorney general can file an enforcement action, the office must give written notice of the violation and, according to the enrolled text's cure provisions, allow an opportunity to cure within a fixed period before proceeding. Confirm the current cure period and any AG rulemaking or complaint-portal procedure directly against the Attorney General's consumer AI rights page, since enforcement infrastructure was still being stood up through 2026.

Civil penalties under Chapter 552 scale with the nature of the violation: a lower range for a violation the court finds curable, a much higher range for one it finds could not have been cured, and a per-day range for a continuing violation. The exact dollar bands are set in the statute; verify the current figures on capitol.texas.gov before citing them in a compliance memo, since they are the kind of number that gets quoted secondhand and drifts.

What TRAIGA does not require

Employers moving to Texas from a state with a more prescriptive AI hiring law should not assume the same obligations travel with them. TRAIGA's employment provision does not require, for a private employer:

• Advance notice to applicants or employees that AI is used in a hiring decision.
• A bias audit, whether internal or independent, or publication of one.
• A documented risk-management program or annual impact assessment.
• Vendor certifications or contractual AI-specific disclosures.

Compare that with Colorado's rules for employers, which from January 1, 2027 require notice and impact assessments for a covered automated decision tool, or NYC Local Law 144, which requires an independent bias audit and public posting. A national hiring process built to Colorado's or New York City's standard will already exceed what Texas requires; the reverse is not true.

Where TRAIGA's disclosure duty actually applies

TRAIGA is not silent on disclosure everywhere; it is silent on disclosure for private-sector hiring specifically. The chapter's disclosure obligations run to two different actors: state agencies that use AI to interact with the public or to make consequential decisions about a person, and health care providers or practitioners using AI in the provision of health care or treatment recommendations. A private staffing agency or in-house recruiting team is neither. If your organization also operates as a state contractor performing a government function, or runs a health care practice with its own hiring pipeline, check whether that side of the business falls under the agency or health care provisions separately from the ordinary hiring use case covered here.

The biometric law sitting next to it

A video interview tool that reads facial geometry or a voiceprint runs into a separate Texas statute, the Capture or Use of Biometric Identifier Act (CUBI). CUBI prohibits capturing an individual's biometric identifier, defined to include a retina or iris scan, fingerprint, voiceprint, or record of hand or face geometry, for a commercial purpose without first informing the person and obtaining consent, and it requires destruction of the data within a reasonable time, no later than one year after the purpose for collecting it ends. The Attorney General enforces CUBI, with civil penalties that can run up to $25,000 per violation; see the Attorney General's CUBI page. TRAIGA amended CUBI to exempt most AI training and development activity from its consent requirement, effective the same January 1, 2026 date, but the consent duty for capturing an individual's own biometric identifier during, for example, a live interview is not itself repealed. If your interview tool analyzes faces or voices rather than only transcribing words, this is the statute to check first, not TRAIGA.

Texas Labor Code Chapter 21: the discrimination law that was already there

Texas Labor Code Chapter 21 is the state's own employment discrimination statute, enforced by the Texas Workforce Commission's Civil Rights Division and closely modeled on Title VII. It applies to an AI hiring tool the same way it applies to a human decision-maker: a selection process, however produced, that has a discriminatory effect or is applied with discriminatory intent can violate Chapter 21 independently of whether TRAIGA's narrower intent standard is met. An employer that clears the TRAIGA bar has not automatically cleared Chapter 21 or Title VII; see Title VII disparate impact and AI hiring tools for how the federal disparate-impact test works on a tool's actual selection rates.

A compliance checklist for Texas employers using AI in hiring

• Confirm which state's AI hiring rules actually apply to each role: TRAIGA reaches AI developed or deployed with discriminatory intent, but Chapter 21 and Title VII reach the outcome regardless of intent.
• Do not skip a bias check just because TRAIGA does not mandate one; run the four-fifths analysis anyway to manage your Chapter 21 and Title VII exposure.
• Identify any interview tool that reads facial expression, facial geometry or voice, and confirm CUBI consent and deletion practices for it, separately from any TRAIGA analysis.
• Keep a record of why each AI hiring tool was selected and configured, since intent is the operative question under TRAIGA itself.
• Do not represent to candidates or clients that Texas "requires" AI disclosure or audits; it does not, and overstating your obligations can create its own problems.
• Watch the Attorney General's site for the complaint and enforcement process as it is built out through 2026.

Common mistakes