Texas Attorney General Ken Paxton has introduced a four-point regulatory framework aimed at AI data centers. The proposal seeks to eliminate Chinese technology from these facilities and establish criminal penalties for AI chatbots that cause harm to children.
A Federal Ban on Chinese Technology in Data Centers
Central to the proposal is a call for federal legislation that would prohibit the use of Chinese technology within AI data centers. According to the report, Ken Paxton is framing this as a critical national security measure to prevent foreign adversaries from gaining a foothold in the United States' critical AI infrastructure.
This move reflects a broader geopolitical trend of "de-risking" from Chinese hardware and software , echoing previous U.S. government actions against firms like Huawei . By targeting the data center level, Ken Paxton is attempting to create a hard perimeter around the physical layer of atrificial intelligence, ensuring that the srevers and networking gear powering next-generation models are not sourced from Beijing.
Criminal Penalties for AI Chatbots that Harm Children
The most controversial aspect of the plan involves the introduction of criminal liability for the facilities that power harmful AI. The report says that Ken Paxton proposes criminal penalties for data centers if the AI chatbots they host are found to have harmed children. This represents a significant shift in how legal liability is viewed in the tech stack, moving the burden from the software developer to the infrastructure provider.
This approach mirrors a growing legislative appetite across various U.S. states to hold tech platforms accountable for the psychological and physical safety of minors. However, by targeting the data center—the physical building and hardware—rather than the AI company itself, the plan introduces a novel and aggressive legal theory regarding the "chain of custody" for digital harm.
Governor Greg Abbott's August 2024 Resource-Limit Policy
While Ken Paxton is not calling for a total moratorium on new AI projects,he has explicitly endorsed the August 2024 initiative launched by Governor Greg Abbott.. This policy allows for a pause in the approval of new data center projects until the operating companies agree to specific limits on their use of local resources, such as water and electricity.
This alignment highlights the tension between Texas's desire to be an AI hub and the reality of its strained power grid. By supporting Governor Greg Abbott's resource-limit policy, Ken Paxton is acknowledging that the massive energy requirements of Large Language Models (LLMs) could potentially destabilize local utilities if left unregulated.
Brendan Steinhauser's Doubts on Enforcement Feasibility
Not all industry observers are convinced the plan is workable. brendan Steinhauser of the Alliance for Secure AI has acknnowledged the political appeal of the proposal but has raised serious questions regarding the feasibility of enforcing the chatbot-related criminal provisions. as reported by the source, the primary concern is how a data center operator can be held legally responsible for the specific outputs of a third-party AI model running on their servers.
This leaves several critical questions unanswered: Would a data center be required to monitor every packet of data to prevent child harm, and how would the law distinguish between the infrastructure provider and the AI developer? Furthermore, the source does not clarify whether these criminal penalties would apply to all data centers or only those receiving specific government subsidies.
Comments 0