The Texas Responsible AI Governance Act: 5 Key Insights
As state legislators across the United States grapple with the complexities of artificial intelligence (AI) regulation, the Texas Responsible AI Governance Act (TRAIGA) emerges as a pioneering piece of legislation poised to reshape the landscape of AI governance. If enacted in its current form, TRAIGA could become the most comprehensive state regulation of its kind.
1. Obligations for Developers and Deployers
TRAIGA mandates that developers and deployers of any AI system that serves as a “contributing factor” to high-risk decisions must adhere to specific obligations. These decisions encompass critical areas such as employment, finance, education, healthcare services, housing, and insurance. Notably, Texas would become the first state to require distributors of high-risk AI systems to exercise “reasonable care” to prevent algorithmic discrimination.
2. Prohibition of Unacceptable Risk AI Systems
The act outlines a clear stance against AI systems deemed to pose an “unacceptable risk.” In alignment with the EU AI Act, TRAIGA would restrict specific uses of AI technology. This includes prohibiting systems that, without explicit consent, are capable of identifying emotions, capturing biometric identifiers, or utilizing such identifiers to categorize consumers based on sensitive attributes.
3. Limited Private Right of Action
While the enforcement of most provisions under TRAIGA would fall to the attorney general, private litigants would retain a limited right of action. Individuals could seek legal recourse for alleged violations involving AI systems that are banned under the legislation, promoting accountability among AI developers and operators.
4. Record-Keeping Requirements for Generative AI
Under the proposed law, developers of generative AI are required to maintain detailed records of the datasets utilized for training their models. This provision aims to enhance transparency and accountability in AI development, ensuring that the sources of training data are well-documented and scrutinized.
5. Exemptions for Small Businesses
TRAIGA recognizes the potential burden that regulations may impose on smaller enterprises. Therefore, the law would exempt many small businesses, defined in accordance with the Small Business Administration criteria, from its obligations. Additionally, developers working on novel AI technologies could apply for exemptions under an experimental sandbox program, encouraging innovation within a controlled environment.
Companies that are investigated under TRAIGA would have a 30-day window to rectify any alleged violations before the attorney general can initiate legal action, fostering a collaborative approach to compliance.
Looking Ahead
The legislative journey of TRAIGA is significant, especially in light of the recent adoption of the Colorado AI Act, which also favored a risk-based governance framework. However, TRAIGA aims to extend beyond Colorado’s measures by offering a broader definition of “high risk,” instituting a ban on certain AI systems, and permitting a narrow private right of action.
As discussions continue, the future of the Texas Responsible AI Governance Act remains uncertain. History suggests that AI oversight bills often undergo revisions that narrow their scope during the legislative process, as seen in the evolution of the Colorado AI Act. Only time will reveal whether TRAIGA will advance in its current form and set a precedent for AI regulation across the nation.
