After months of ongoing discussions and questions regarding how the state will approach artificial intelligence, a state representative finally filed House Bill 1709 -- the proposed Texas Responsible Artificial Intelligence Governance Act -- on Monday.
State Rep. Giovanni Capriglione (R-Southlake) initially released a draft of the proposed act in October and officially filed the legislation on Monday and will likely introduce it in the upcoming legislative session, which is set to begin on Jan. 14. Capriglione co-chairs the state's AI Advisory Council.
In contrast to traditional computer programs, which follow predetermined instructions, AI technologies have the potential to learn from new data and change over time, as humans do.
The 44-page act would establish regulations, guidelines and obligations for the development, use and distribution of "high-risk AI systems."
Major takeaways from the act:
"Artificial intelligence continues to increasingly reshape our daily lives, and we must approach this transformation with necessary care," Capriglione wrote in a release about the filing Monday evening. "The Texas Responsible Artificial Intelligence Governance Act ensures we embrace innovation while safeguarding public trust, privacy, and safety. By filing this legislation, Texas is taking a proactive approach to addressing the challenges and opportunities of AI, ensuring that Texas remains a hub for responsible technological advancement."
Along with other regulations and guidelines, the bill would prohibit certain AI systems that pose "unacceptable risks,": manipulate human behavior, engage in social scoring, capture biometric identifiers, infer or interpret sensitive personal attributes, infer emotions without consent or produce deepfakes that constitute child sexual abuse material or intimate imagery prohibited under state law.
Two Texas mothers filed a lawsuit earlier this month against Character.AI -- a generative AI companion chatbot that allows users to have conversations with bots posing as celebrities or characters of their choosing -- for promoting sexually explicit content to children and encouraging self-harm and violence.
After the filing, Capriglione told the American-Statesman that his then-incoming bill would protect consumers like the two mothers and their children and keep companies from developing tech that "knowingly manipulate or a person's behavior in a manner that is likely to cause harm."
The bill would require developers of generative AI -- a type of AI that uses machine learning to create new content to act and feel more human-like -- to keep "detailed records" of training datasets.
"AI innovation in Texas can undeniably flourish without relying on applications that target and negatively impact children," Capriglione said in an email statement following the lawsuit filing this month.
The act would also create an "AI Regulatory Sandbox Program" for participating developers to test AI systems. Those in the sandbox program would be exempt from certain aspects of the act to promote research, training, testing and pre-deployment activities within the scope of the program.
The act would authorize the Texas attorney general to enforce the requirements and regulations.
The act also proposes amendments to Texas' Data Privacy and Security Act, also written by Capriglione, to incorporate AI-specific regulations and language as well as providing an AI workforce grant program and a new "AI council" to provide advisory guidance.
With the passage of the Texas' Data Privacy and Security Act, Texas became the 10th state to pass comprehensive data privacy legislation last year. The Texas Data Privacy and Security Act gives Texans the right to prevent companies from collecting or selling their data. It will also allow residents to request copies of their personal data; give them the ability to correct data; and forbid businesses from discriminating against those who exercise these rights.
Other exemptions from the act beyond those in the sandbox program would include small businesses and developers of open-source AI systems if the developer has taken action to prevent high-risk use and makes the "weights and technical architecture" of the AI system publicly available.
Following the lead of several Democratic-led states, the act is modeled after the Colorado AI Act, which was passed in May. If enacted, Texas would be the second state to have an industry-agnostic and risk-based piece of AI legislation.
California Gov. Gavin Newsom vetoed a proposed AI safety bill in his state after it generated strong opposition from major technology companies in Silicon Valley, many of which claimed the bill would have hindered innovation and stifled growth as companies compete in the global AI race. Colorado's governor has already asked for amendments to its AI regulatory law before it goes into effect in 2026.
"HB 1709 reflects extensive consultation with members of the public, industry leaders, academics, and advocacy groups," Capriglione said in the Monday release. "By balancing innovation with public interest, we aim to create a blueprint for responsible AI use that other states and nations can follow. Texas has always been at the forefront of technological progress, and with this bill, we are ensuring that progress is ethical and beneficial to all Texans."