The United States saw significant regulatory activity around agentic AI this week, with Virginia and Texas leading state-level initiatives. Virginia Governor Glenn Youngkin launched a groundbreaking pilot program using agentic AI to analyze state regulations. This system will autonomously scan Virginia's regulatory documents to identify contradictions, redundancies, and outdated language. The project builds on Virginia's earlier success in reducing regulatory requirements by 26.8% and cutting guidance document text by nearly half. Governor Youngkin stated this AI-powered review will "unleash Virginia's economy" by removing bureaucratic barriers.

Texas implemented the Texas Responsible AI Governance Act (TRAIGA), establishing comprehensive rules for AI developers and users. The law mandates clear disclosures when individuals interact with AI systems in government services or healthcare, requiring notices even when the artificial nature might seem obvious. TRAIGA also prohibits AI-generated non-consensual pornography and child exploitation content while modifying Texas' biometric privacy law. Notably, it allows companies to use publicly available biometric data for AI training without consent, unless specifically identifying individuals.

Across the U.S., state legislatures have introduced over 1,000 AI-related bills in 2025 alone. This follows 2024's nearly 700 proposed bills across 45 states. Current proposals fall into four main categories: transparency requirements (like California's upcoming training data disclosure rules), algorithmic fairness (modeled after Colorado's law), safety protocols (including proposed emergency shutdown mechanisms), and accountability frameworks. Virginia's governor recently vetoed algorithmic fairness legislation, highlighting policy disagreements between states.

At the federal level, the Trump administration proposed a 10-year moratorium on state AI regulations through the "Big Beautiful Bill". Supporters argue this would prevent a patchwork of conflicting state laws that could burden interstate commerce. Critics counter that states are filling a regulatory void, with Texas' TRAIGA taking effect before Colorado's similar law next year.

Globally, the EU AI Act is driving discussions about documentation standards for agentic systems. Unlike traditional AI, agentic systems autonomously set goals, interact with tools, and adapt behavior post-deployment. The EU's requirement for "comprehensive technical documentation" presents challenges since current methods track static systems, not evolving agents. Experts emphasize that agentic AI needs continuous "living documentation" to meet regulatory demands for transparency and accountability.