
Imagine a fintech startup that deploys an artificial intelligence (“AI”) model to underwrite small-business loans—only to face a demand letter alleging “intentional discrimination” under Texas law. On June 22, 2025, Governor Greg Abbott signed the Texas Responsible Artificial Intelligence Governance Act (“TRAIGA”), placing Texas at the vanguard of state-level AI regulation. As the fourth state to enact a comprehensive AI statute—after Colorado, Utah, and California—Texas now offers both a clear roadmap for developers and heightened risks for those whose AI decisions cause harm.
With the passage of TRAIGA, business trial lawyers are entering an era where AI regulation is no longer a theoretical concern but a live, litigable issue. For trial attorneys handling commercial disputes, TRAIGA’s unique blend of intent-based liability and centralized enforcement reshapes the evidentiary landscape, requiring more rigorous documentation and strategic foresight. Business lawyers must now anticipate how a client’s AI-related decisions prone to allegations of bias—especially in areas such as lending, hiring, and marketing—might be scrutinized under a standard that demands proof of knowing and intentional misuse.
TRAIGA, which takes effect January 1, 2026, introduces a suite of targeted prohibitions, regulatory mechanisms, and compliance frameworks that will shape not only the development and deployment of AI systems in Texas but also the broader landscape of business litigation and regulatory strategy. For business trial lawyers, understanding the contours of this new law is essential, as it will influence litigation strategy, evidentiary standards, and the future of AI-driven business operations.
The version of TRAIGA that was passed and signed into law represents a pared-down evolution from earlier, more expansive drafts that mirrored the risk-based frameworks of the EU AI Act and the Colorado AI Act. The final version, however, reflects a pragmatic shift toward outcome-focused regulation, emphasizing specific prohibited uses of AI while scaling back broad compliance mandates for the private sector. This approach has direct and nuanced implications for business litigation in Texas and potentially beyond.
Intent-Based Evidentiary Standard for Liability
Under Section 4 of TRIAGA, civil liability attaches only where a developer or deployer “intentionally” uses AI to
- promote self-harm or suicide;
- promote harming another person;
- facilitate criminal activity;
- engage in unlawful discrimination;
- produce unlawful deepfakes or child-exploitation content; or
- infringe, restrict, or impair constitutional rights.
The elevated burden of proof placed on plaintiffs is one of TRAIGA’s most consequential implications for business trial attorneys. This intent-based standard departs from risk-only frameworks (e.g., the EU AI Act), requiring claimants to adduce evidence of purposeful misconduct—not merely disparate outcomes. Traditional arguments relying on disparate outcomes or statistical disparities will no longer suffice without evidence of deliberate intent; claimants must now demonstrate that the developer or deployer acted to discriminate or cause harm intentionally.
This heightened standard may reduce the volume of AI-related business litigation premised on algorithmic bias, particularly in employment, lending, and other regulated sectors. It also makes early discovery strategy and internal documentation absolutely pivotal, placing a premium on robust documentation and internal controls. TRAIGA’s safe harbor provisions—such as affirmative defenses for parties that discover violations through internal review, adversarial testing, or compliance with recognized risk management frameworks (e.g., the National Institute of Standards and Technology (“NIST”) AI Risk Management Framework)—may encourage businesses to adopt proactive compliance measures and self-audit protocols. Attorneys representing businesses should proactively advise clients to develop and maintain records showing compliance with these recognized frameworks to activate the safe harbor protections and blunt any allegation of purposeful harm. Further, businesses should strengthen internal controls and preserve audit trails demonstrating their AI systems’ legitimate aims and operational safeguards. These records can become powerful tools during motions to dismiss or summary judgment.
TRAIGA vests exclusive enforcement authority in the Texas attorney general, precluding private rights of action. The attorney general is empowered to investigate alleged violations, issue civil investigative demands, and seek injunctive relief and civil penalties ranging from $10,000 to $200,000 per violation, with additional penalties for continuing violations. TRAIGA also provides for a sixty-day cure period following notice of a violation, incentivizing prompt remediation. This centralized enforcement model may streamline the adjudication of AI-related disputes, reduce the risk of inconsistent outcomes, and provide greater predictability for businesses.
Impact on Discovery and Pretrial Practice
TRAIGA’s focus on intent and its explicit exclusion of liability for unintended third-party misuse of AI systems may limit the scope of discovery into downstream uses. In traditional business litigation, especially in cases involving products or technologies with broad downstream applications, plaintiffs often seek extensive discovery into how a product was used by third parties, the range of possible outcomes, and the foreseeability of misuse. Such discovery can be costly, time-consuming, and burdensome, as it may require the production and analysis of voluminous data regarding end-user behavior, system outputs in varied contexts, and communications with customers or partners.
AI tools like Technology-Assisted Review (“TAR”) can bring order to an overwhelming amount of complex data that in recent years had complicated discovery and pretrial motion practice while creating a massive litigation expense burden. TRAIGA clarified the lines as to what behavior is culpable, which reduces disputes over the adequacy of internal controls and the potentially limitless downstream effects of AI systems in the hands of third parties. And because TRAIGA excluded TAR from its scope, it implicitly affirms continued use of TAR to streamline discovery of just the relevant evidence.
Potential Benefits and Drawbacks of TRAIGA’s Sandbox Regulatory Model
One of TRAIGA’s most innovative features is the establishment of a regulatory sandbox program, administered by the Texas Department of Information Resources in consultation with the Texas Artificial Intelligence Council established by the legislation. The sandbox allows approved participants to develop and test AI systems in a controlled environment, temporarily exempt from certain state licensing and regulatory requirements, for up to thirty-six months.
The sandbox model offers regulatory flexibility and innovation by providing a structured pathway for businesses to experiment with novel AI applications without the immediate risk of regulatory penalties, which is particularly valuable in areas of legal uncertainty where traditional regulatory frameworks may lag behind technological advances. Additionally, by requiring quarterly reports on system performance, risk mitigation, and stakeholder feedback, the sandbox generates empirical data that can inform future legislative and regulatory reforms, enhancing the capacity of regulators and lawmakers to craft targeted, effective policies. Moreover, the Texas sandbox positions the state as a leader in AI regulation, potentially serving as a model for other jurisdictions and facilitating cross-jurisdictional data analysis, which may lay the groundwork for future reciprocity agreements, enabling businesses to scale innovative AI solutions across state lines with greater legal certainty. For business law practitioners, the sandbox offers a clear, time-limited framework for managing regulatory risk during the development and deployment of cutting-edge AI systems, and participation in the sandbox may also serve as a mitigating factor in enforcement actions, further incentivizing compliance.
However, the sandbox model also presents potential drawbacks. While it promotes innovation, it might contribute to a fragmented regulatory landscape if other states adopt divergent models or standards, creating challenges to harmonizing compliance across jurisdictions for businesses operating nationally. Effective oversight of the sandbox requires significant administrative resources, including technical expertise and ongoing monitoring, potentially creating barriers to entry for smaller businesses and limiting the program’s inclusivity. While the future of state-level AI sandboxes is safe for the moment, it remains uncertain due to the prospect of federal preemption. Though one had been proposed, there was no state AI law moratorium in the recently signed federal budget bill; it was stripped out of the U.S. House version by the U.S. Senate. Still, the political forces that had it included in the first place may try again. This could suspend or nullify the Texas sandbox and similar programs, making it essential for business law professionals to closely monitor federal developments, as the regulatory environment may shift.
Practical Advice for Business Law Practitioners
Ultimately, TRAIGA isn’t just a compliance statute—it’s a blueprint for how AI liability will be litigated. For business trial lawyers, this signals a shift in how risk is assessed, evidence is preserved, and cases are pled. Those who understand TRAIGA’s enforcement structure, sandbox incentives, and documentation expectations will not only better defend their clients but also shape emerging jurisprudence on AI accountability. In this evolving landscape, the ability to speak the language of both law and machine will become a key differentiator in the courtroom.
TRAIGA marks a watershed moment in the evolution of AI regulation. By focusing on specific, outcome-based prohibitions and embracing innovative regulatory mechanisms such as the sandbox model, Texas has crafted a framework that attempts to balance the imperatives of innovation, consumer protection, and legal certainty.
For business law practitioners, TRAIGA’s intent-based liability standard and robust safe harbor provisions offer both challenges and opportunities. Practitioners should consider advising clients to do the following:
- Maintain clear records of the intended uses and operational controls of AI systems to support defenses against claims of intentional misconduct.
- Implement and document compliance with frameworks such as the NIST AI Risk Management Framework to avail themselves of statutory safe harbors.
- Stay abreast of federal and state legislative activity, particularly regarding potential preemption and the evolution of sandbox programs.
- Evaluate whether to join the regulatory sandbox for innovative AI projects, balancing the opportunities for experimentation against the administrative requirements and potential for regulatory change.










