This weekly update reveals significant developments in how governments worldwide are preparing to regulate AI agents - sophisticated computer programs capable of autonomous decision-making and planning.

Italy achieved a historic milestone by becoming the first European Union member state to enact comprehensive national legislation specifically addressing AI agents and their deployment. Italy's Law No. 132 of 2025 entered into force on October 10, 2025, creating a pioneering framework that complements the broader EU AI Act. This legislation represents a crucial step forward in establishing clear rules for how AI agents can operate within sensitive sectors including healthcare, public administration, judicial systems, and employment.

The Italian approach emphasizes human supervision as a cornerstone principle for AI agent deployment. In healthcare settings, the law permits AI agents to function as support tools for medical professionals but explicitly prohibits their use in making final treatment decisions or discriminating in patient care access. This balanced approach recognizes the potential benefits of AI agents while maintaining essential human oversight in life-critical situations. The law also establishes robust protection for minors through a dual-tier consent framework, requiring parental approval for AI agent interactions with children under fourteen.

Italy designated the Italian Digital Agency (AgID) as the notifying authority responsible for accrediting bodies that verify AI system compliance, while the National Cybersecurity Agency (ACN) serves as the primary supervisory authority for enforcement and sanctions. However, this governmental approach has raised concerns about regulatory independence, as the European Commission previously emphasized the need for independent oversight bodies in AI regulation.

In the United States, regulatory approaches to AI agents remain fragmented across state lines. During 2025, lawmakers tracked 210 AI-related bills across 42 states, with only 20 achieving enactment - representing a mere 9% success rate. Most significantly for AI agents, legislators began explicitly recognizing "agentic AI" as a distinct category requiring specialized governance approaches. States like Virginia introduced "regulatory reduction pilots" while Delaware established agentic AI sandboxes, representing early experiments in managing autonomous AI systems.

The challenge with AI agents lies in their autonomous planning capabilities that extend far beyond traditional AI applications. Unlike generative AI systems that simply create content, agentic AI can develop complex strategies, make sequential decisions, and adapt their behavior based on changing circumstances. This creates unprecedented regulatory complexity because traditional risk frameworks struggle to address systems capable of independent reasoning and action.

Regulatory experts highlight that existing compliance models may prove inadequate for agentic AI systems. Traditional AI regulations focus on specific use cases or risk levels, but AI agents can dynamically shift between different functions and risk categories during operation. When an AI agent makes multiple interconnected decisions to achieve a goal, determining accountability for negative outcomes becomes significantly more complex than with single-purpose AI tools.

International coordination emerges as a critical necessity given the global nature of AI agent deployment. The European Union's AI Act provides extraterritorial reach, affecting U.S. businesses that develop or deploy AI solutions targeting European markets. This creates a complex compliance landscape where companies must navigate multiple jurisdictional requirements simultaneously, particularly challenging for AI agents that may operate across borders or serve international user bases.

Looking ahead, the regulatory landscape for AI agents will likely require adaptive governance frameworks that can evolve alongside rapidly advancing technology. Traditional static regulations may prove insufficient for systems capable of learning and modifying their own behavior. Regulators must balance fostering innovation with protecting public safety, requiring unprecedented cooperation between technologists, policymakers, and international bodies to establish effective oversight mechanisms for the autonomous AI future.