What Texas Employers Need to Know About AI and the Law
Texas employers have been enthusiastically adopting artificial intelligence tools to screen resumes, score candidates, schedule interviews, and evaluate performance. The appeal is obvious: AI is fast, consistent, and tireless in ways that no HR department can match. The legal exposure, however, is keeping pace with the technology in ways that many employers have not yet internalized. Texas now has its first AI governance law on the books, the federal regulatory picture has shifted dramatically, and a California courtroom is quietly rewriting what it means to be liable for a hiring decision you did not technically make yourself. If the algorithm is doing your hiring, you still own the outcome.
The Texas Responsible Artificial Intelligence Governance Act
In June 2025, Governor Greg Abbott signed the Texas Responsible Artificial Intelligence Governance Act, known as TRAIGA (HB 149), into law. TRAIGA became effective on January 1, 2026, making Texas the third state, after Colorado and Utah, to pass legislation specifically governing artificial intelligence systems.Employers who tracked this legislation since its aggressive introduction as HB 1709 in December 2024 are entitled to exhale, because the final law is considerably narrower than what the original bill proposed.
The original version, TRAIGA 1.0, would have required private employers to conduct mandatory bias audits, disclose AI use to job applicants and current employees, and maintain detailed risk management policies. However, the enacted version of TRAIGA does none of those things. There are no mandatory impact assessments, no bias testing requirements, no requirement for private employers to tell applicants or employees that AI is being used in decisions about them, and no obligation to implement a risk management AI policy. Enforcement authority rests exclusively with the Texas Attorney General, and the law contains no private right of action for an individual to sue for an alleged TRAIGA violation. In other words, individuals (including employees or job applicants) may submit their complaints to the Texas Attorney General but only the Texas Attorney General can make the decision to sue for an alleged TRAIGA violation.The law also includes a 60-day cure period, meaning employers who receive notice of a violation have time to remediate before facing penalties.
What TRAIGA does prohibit, for both private and public employers, is the development or deployment of an AI system with the intent to discriminate against a protected class under state or federal law. In other words, employers cannot utilize AI to perform actions the employer would have been prohibited from doing themselves. For purposes of determining whether there is a TRAIGA violation, proof of intentional discrimination is required and disparate impact alone is not enough. This standard aligns with Executive Order 14281 and its preference for intent-based over impact-based enforcement.
However, it is important for employers to still be aware of the potential disparate impacts their utilization of AI may be creating. Employees, job applicants, and federal agencies like the EEOC can still proceed under a disparate impact theory even if the Texas Attorney General cannot bring a TRAIGA claim.
TRAIGA is, in short, a regulatory reprieve. The question is whether employers treat it as a finish line or a head start.
The Federal Landscape Shifted, But the Law Did Not
Executive Order 14179 of January 23, 2025 “Removing Barriers to American Leadership in Artificial Intelligence” broadly targeted the previous administrations AI policies by labeling them as “barriers to American AI innovation” and directed federal agencies to suspend, revise, or rescind prior AI policies. As a result, In federal agencies subsequently removed AI-related guidance to include the EEOC’s 2023 technical assistance document (which applied Title VII’s disparate impact framework to algorithmic hiring tools) and the Department of Labor’s AI best practices guidance were both removed. This led to confusion amongst employers, with many employers interpreting those removals as a green light to use AI in ways that violates Title VII of the Civil Rights Act of 1964, the Age Discrimination in Employment Act, and the Americans with Disabilities Act (collectively “Federal Employee Protections”).
Executive Order 14179, however, does not give employers the green light to use AI to violate Federal Employee Protections. Not only do these Federal Employee Protections remain fully in effect they also apply to the use of AI in employment decisions in exactly the same way they apply to any other selection procedure. For example, the EEOC’s prior published AI guidance did not create any new obligations. Rather, this guidance explained how the existing laws already covered AI. The removal of this guidance from the websites did not remove the underlying law which this guidance was explaining.
In other words, Executive Order 14179 should not be taken by employers as a sign they can begin using AI recklessly. Employers who use AI tools that produce discriminatory outcomes still remain exposed under federal civil rights statutes, regardless of the current administration’s enforcement priorities. Prior EEOC statements have made abundantly clear that unlawful discrimination can be intentional(such as programming a resume screener to reject applicants based on a protected characteristic)or unintentional (when a neutral tool produces a disparate impact without job-related justification). If an employer is utilizing AI for screening job candidates or making decisions about employees, the employer should have a process in place to ensure the employer’s use of AI is not unknowingly exposing the employer to liability.
The Vendor Problem Nobody Wants to Talk About
Here is where things get genuinely uncomfortable for employers who assumed that outsourcing hiring decisions to a third-party AI platform also outsourced their legal exposure.
In July 2024, the United States District Court for the Northern District of California issued a ruling in Mobley v. Workday, Inc., 740 F. Supp. 3d 796 (2024), holding that a third-party AI hiring software vendor can be named as a defendant in employment discrimination lawsuits under an “agent” theory of liability. Derek Mobley, a Black man over 40 with anxiety and depression, applied to more than 100 positions at companies using Workday’s AI-based screening tools and was rejected every time, in some cases within an hour of submission. He alleged that Workday’s algorithms discriminated based on race, age, and disability in violation of Title VII, the ADEA, and the ADA.
The court found that Workday’s platform was not simply implementing criteria that employers set, but was instead participating in the decision-making process—recommending some candidates to advance and rejecting others—which was sufficient to allow the agent liability claim to proceed. In May 2025, the court conditionally certified a collective action under the ADEA, potentially extending the case to a large number of applicants processed through Workday’s system since September 2020. Workday represented in filings that 1.1 billion applications were rejected using its software during the relevant period.
The takeaway for employers is not that AI hiring tools are unusable. It is that using an AI vendor does not transfer your discrimination liability to that vendor. Employers should make sure they are able to articulate how these AI hiring tools are utilized and what control systems employers have in place around these AI hiring tools. Courts have consistently held that employers cannot escape liability for discrimination by delegating traditional employment functions to a third party, whether that third party is human or algorithmic. The Workday case did not invent a new principle; it applied an existing one to new facts.
What Employers Should Be Doing Right Now
The combination of TRAIGA’s passage, the federal guidance pullback, and the expanding Workday litigation points toward a clear set of practical actions for employers. These are not hypothetical future considerations. They are current-state risk management.
Employers should audit their AI tools before a court does it for them. If AI is involved in any employment decision(from recruiting through termination),employers need to document what the tool does, how it was selected, and what testing the vendor has performed for bias. Employers should also go a step further and build out policies and procedures for how AI tools should be utilized and what checks are in place for these AI tools. Employers should ask vendors directly whether their tools have been evaluated for disparate impact on race, age, and disability, and request documentation. A vendor’s assurance that its tool is bias-free does not protect an employer from liability if that assurance turns out to be wrong but it demonstrates that an employer was performing due diligence activities to investigate an AI tool prior to incorporating it into the employer’s system.
When performing self-audits, employers should apply the four-fifths rule as a benchmark. Under the Uniform Guidelines on Employee Selection Procedures, adverse impact is generally indicated when a protected group’s selection rate is less than four-fifths (80%) of the highest-performing group’s selection rate. Running this analysis periodically on AI-assisted hiring outcomes is not a guarantee of compliance, but it is a reasonable indicator of whether a problem may be developing so you do not have to wait for a legal problem to occur before you can begin remedial actions.
Employers should also regularly review their vendor contracts. If an AI hiring vendor is found to be acting as an employer’s agent for liability purposes, the terms of this contract will matter. Indemnification provisions, audit rights, and representations about nondiscrimination testing are not boilerplate; they are leverage. This is a contract employers should make sure to review with counsel prior to signing and prior to automatic renewal dates.
Most importantly, employers should maintain meaningful human review. A human decision-maker who actually reviews AI-generated candidate rankings and makes the final call is not a complete defense, but it creates a materially different legal posture than a system that auto-rejects applicants without any human involvement. The Workday court specifically noted rapid rejection timing as support for the inference that automated systems were making decisions. Human oversight creates a record of considered judgment.
The Bottom Line
TRAIGA is genuinely good news for Texas employers compared to what the original bill proposed. But “less regulated than feared” is not the same as “unregulated.” Federal civil rights law still applies to every AI tool in your stack, private litigation is filling enforcement gaps, and the Workday case previews what happens when employers assume that algorithmic distance from a hiring decision is the same as legal distance.
Treaty Oak Employers’ Law Group counsels employers across Texas, Wyoming, Colorado, and North Carolina on the practical intersection of employment law and technology-assisted decision-making. If artificial intelligence is part of how your organization hires, manages, evaluates, or terminates employees, now is the time to understand how those tools are operating, what controls are in place, and where legal exposure may exist.
To take a closer look at your organization’s AI-related employment risks, schedule a consultation with Treaty Oak Employers’ Law Group.